Determining Electron Temperature and Plasma Potential
An Additional Concern for Temperature Analysis
This is a simple example of Langmuir probe analysis and the issues related to it. It is intended to serve as a helpful reminder of the technical details in analyzing the data from a swept Langmuir probe and is not a complete theoretical effort. If you have already familiarized yourself with Langmuir probe theory, then you may find this treatment helpful. In this example I begin with the data acquired by measuring the current drawn by a Langmuir probe as the bias applied to that probe is varied. This data is analyzed in order to determine the plasma density, temperature, and potential. While the concept of Langmuir probe usage, digitization of the received data, and even the engineering of the diagnostic are all worthy of discussion, the following has a focus on analysis in order to limit the size of this entry. The data presented in this example was obtained in the undergraduate plasma laboratory (PHYS 180E) at UCLA while I was testing the equipment and helping in the design of the lab exercise as a teaching assistant for the course. Figure 1 represents the circuit used to acquire the signals that will be processed.
An example of what the raw data may look like is provided in figure 2. This data is obtained from an oscilloscope that averages 16 separate acquisitions (while the plasma is continuous, CW, the probe bias sweep is made at a rate of 4 Hz, thereby allowing for multiple acquisitions to be averaged over for one final result). The x-axis units represent the data point number (i.e. if this data was in a spreadsheet, then data point 1000 is the one-thousandth data point you have). The trace labeled Vbias represents the applied voltage to the probe. For a properly set up oscilloscope, this signal will be output in the correct units and calibration. The other trace, VR, represents the voltage measured across a resistor in series with the probe. For a given resistor, R, the current through it, Iprobe (named because it is also passing through the probe) is found using V = IprobeR.
Notice that the plot in figure 2 shows a transient effect in the VR trace for values near data index of zero. Be sure to extract only the “proper” portion of these traces when performing your analysis. Zooming in on the data will reveal that we should only consider points 20 through 2500. The points prior to number 20 are an artifact of the bias voltage turn-on. Another striking characteristic of the plot is the stepping feature of VR and to a lesser extent also of Vbias. This is not a plasma physics result, rather, the Vbias power supply achieves its sweep by stepping the potential up over time. This sweeper is typically operated at a frequency in the kilohertz range and the steps are very difficult to observe. In the 180E lab the plasma is steady state and the sweep frequency has been lowered as much as possible. The sweep rate used is 4 Hz, which results in the individual steps being more noticeable.
It is necessary to convert the VR signal into units of current. Since the electron temperature, Te, trace involves derivatives of the current we must process these raw data in multiple ways. To convert VR into Iprobe we can use Iprobe = VR/R. For a resistor of R = 677 Ω this becomes Iprobe = 1.4771 × 10-3 VR.
Processing the Raw Data
Figure 3 presents a first look at the actual IV trace. This is probe current plotted as a function of probe bias. This plot can tell us a lot about the plasma. The floating potential, Vf, occurs where Iprobe = 0, which appears to give Vf ≈ -40 V. The ion saturation current, Isat, is seen at biases well below Vf. The roll-off at large positive values of Vbias (technically, in this case the largest values of Vbias approach zero and may not actually go positive) corresponds to the electron saturation current, esat. The location of this roll-off, or knee, is the plasma potential, Φp. As with the floating potential, some estimate of this value may be made from the plot, but a more accurate method will be used to determine the final value. Figure 3 displays Φp ≈ -15 V.
For swept Langmuir traces such as those presented here, the most important relation between the measurement and plasma parameters is given by,
where q is the electron charge, kB is Boltzmann’s constant, and the constant term will not be important. Notice that this relationship is in the form of a line given by a function f such that f(V) = mx + b, where m is the slope of the line and b is the y-intercept. If our x is actually x = Vbias – Vf, then the slope of this line is related to the electron temperature. Our method is to plot the term ln|Iprobe – Isat| and then fit a line to it. The slope of this best fit is inversely proportional to the electron temperature.
It is necessary to subtract the value of the ion saturation current from Iprobe in order to continue with the analysis. A closer look at figure 3 provides Isat = 1.73 × 10−4 A. It is important to get an accurate value of Isat so be sure to read its value careful (i.e. not from a wide range plot like that shown in figure 3). Since the Isat value is negative, subtracting it from Iprobe will result in almost all of the values of Iprobe − Isat being positive. This is the intention because we will be working with the natural logarithm of the current in the next few steps and the natural log is not defined for negative numbers. If you have a few negative values left in your trace after subtracting the Isat value that is acceptable. Those values will not play a role in the temperature calculation that follows. Figure 4 shows the resultant electron current with respect to the total current.
Determining Electron Temperature and Plasma Potential
Figure 5 represents the logarithmic plotting of the electron current. The electron current is Iprobe − Isat because we have removed the ion contribution to the total current by subtracting Isat. The plot decays very rapidly for values of Vbias < Vf. This is because ln(0) = -∞ and by subtracting the value of Isat we have forced the current trace to be near zero for values of Vbias < Vf. The part of this trace that we are interested in occurs above the floating potential and this is the region presented in the next plot.
Figure 6 is a zoomed in version of figure 5 that also includes linear fits to the electron saturation current and inverse temperature. The temperature fit is performed over the range −22 < Vbias < -17 V, which can be seen as the exponentially rising region in figure 4. It is important to choose a bias range over which to perform this fit that represents the temperature dependent increase in probe current. By overplotting your data with the linear fit it is possible to quickly demonstrate whether the correct temperature region has been identified. According to this temperature fit, the inverse temperature is approximately 0.27, which leads to Te = 3.70 eV.
A linear fit to the electron saturation current is also shown in figure 6. The intersection between this fit and the temperature fit occurs at the plasma potential. This results in a reading of Φp = -14.3 V.
Calculating Electron Density
Knowing both the electron temperature and the ion saturation current allows us to calculate the electron density using,
where the new terms are M for the ion mass and As which represents the area of the probe sheath. For cases in which the applied probe bias does not greatly exceed the value needed to obtain one of the saturation currents we may approximate the sheath area as the probe tip area. For significant overbiasing this is not a good approximation. In most cases this criteria is met and the approximation is one of the smaller sources of error for probe measurements. The concepts of sheath expansion and Debye length with respect to probe size are worthy of discussion in another effort.
For argon we have M = 6.62 × 10−26 kg. The planar probe used to collect this data has an area of Aprobe = 0.738 cm2 (this is the area of one side multiplied by two because it collects ions and electrons from both faces).
To calculate the electron density it is possible to directly insert the electron temperature in units of electron-Volts by noting the following relationship,
where on the left side the temperature is in units of Kelvin.
For the temperature measured in this setup, Te = 3.70 eV, the electron density is found to be ne = 8.09 × 1015 particles per cubic meter. Most plasma physicists use cgs units and would report this as a density of 8.09 × 109 cm-3.
An Additional Concern for Temperature Analysis
This is a treatment to account for the presence of a hot electron distribution in the plasma. For plasmas that are generated through a fast electron breakdown process, such as an electron gun or cathode-anode pair, this effect is likely relevant.
In the preceding section the properties of our plasma have been determined under a few assumptions. Possibly the most significant assumption was that the plasma source (i.e. electron beam or electron spray) did not affect the IV trace. This is not absolutely correct and it is possible to include effects of the plasma source in our interpretation of the data. From a conceptual standpoint, we may expect that our system contains two separate electron populations. One population is the background plasma (hence referred to as the plasma electrons) and the other is from the plasma source directly (beam electrons). Since the plasma electrons are generated by ionization from collisions between beam electrons and background neutral gas they cannot possibly be more energetic than the beam electrons. This translates into the plasma electrons having less energy than the beam electrons, which is equivalent to them having a lower temperature, Te,plasma < Te,beam.
If there are two separate electron populations in the system, then it may further be expected that the IV trace displays two separate linear regions when plotted in the ln|Iprobe – Isat| fashion. In order to discern these regions it is necessary to better define the behavior that corresponds to the plasma temperature. If there is a hot electron tail (referred to as a tail because this represents the tail end of the electron velocity, or energy, distribution function), then those electrons will continue to strike the probe even after we have made Vbias more negative. If these electrons are affecting the probe measurement, then we should expect to see a second linear region in the logarithmic plot for values of Vbias approaching the floating potential. As previously, we must avoid actually considering the floating potential because that is where the Isat effects have been subtracted to return a near zero electron current value.
Figure 7 shows a fit to the tail electron region of the curve. The slope of this line is the inverse of the hot electron temperature, Te,hot = Te,beam. The slope of this line (the blue line in figure 7) is less than that of the plasma temperature, but since the slope represent the inverse electron temperature this correctly gives a hotter temperature for these tail electrons. The fit gives Te,hot ≈ 10 eV. The fit is made over the range −36 < Vbias < −22. A slightly different range may have provided a better fit. It is always a challenge to verify that you have made the best possible fit. An inclusion of error analysis and a quantitative measure of the quality of these linear fits will help justify your reported values.
It is not correct to use this temperature to calculate the density and plasma potential of the beam electrons. While they do skew the IV trace, we are still applying the general theory of quasi-neutrality and ion collection to the interpretation of the IV characteristic. There are no ions included in the beam distribution and the rest of our probe theory does not apply.
Conceptually, the claim is that the probe collects some of these beam electrons and treats them as plasma electrons. The probe cannot possibly know the difference between the different electrons it collects. To the extent that this simplified notion is true, the beam electrons result in an IV trace that returns a larger temperature than is correct for the plasma that the beam created.
Sample Data for Practice
A spreadsheet file is available for download at the following link:
Sample Langmuir Data
As an actual data set this file contains noise and error, both of which contribute to the authentic experience of analysis. The values of plasma temperature, energetic tail temperature, floating potential, and ion saturation current that I obtained are included in the file for comparison to your own results. It is certainly possible to get a better (i.e. more accurate) result than I did.
If you are interested, then good luck.
Alex Murphy says
Hi Dr. Pace,
Thank you for the excellent explanation! I am trying to curve fit triple Langmuir probe data and obtain the plasma temperature and density from this curve fitted function. I have three data points to fit with: a point in the ion saturation region, the floating potential, and a point in the electron saturation region. I have tried curve fitting the data to a sigmoidal function and taking the slope in the electron saturation to obtain the electron temperature region but using your data this gives Te=126.94 eV. Do you have any suggestions for how to get a more accurate fit? Are there alternative analysis methods I should be considering? The main issue is that I must assume a thick sheath plasma and as such I cannot use many of the simplifications for triple Langmuir probe data analysis. Thank you! [email removed]
I don’t quite understand, the benefit of the triple probe technique is that you can avoid performing fits, e.g. the triple probe section here,
Sorry, don’t think I can provide any guidance here.
Thanks for your article.. Can you suggest how to get Ion saturation current from IV data.
Sure thing: the ion saturation is the current that you draw while the probe is biased sufficiently negative. If you check out Figure 3, you’ll see that the current is nearly fixed as the probe bias moves further negative. The ion current has “saturated,” which is the value you seek. If you bias too far negative, then other effects come into play and the current will rise further, so it helps to review the curve to identify the true saturation region.
Hope this helps,
Anmol Taploo says
Hi Dr. Pace,
I am a PhD student. I tried obtaining Te from the sample data you have posted but I got value around 4.13 eV instead of value you have mentioned. Is this the correct value. Also, do you have any additional sample data that I can try using to practice,
That value is pretty close, is it possible that some of the data values were rounded early in the processing? Additionally, if you perform the linear fits across a different range, then you could get a different final answer (not clear whether it would be this different). This is the only data set I have, but I think there might be other data in the PlasmaPy distribution or from other plasma physics groups.
Best of luck in your studies,
Anmol Taploo says
Thanks, do you have data set for figure 7 if possible. I can take a look. Thanks. Also, when plotting ln(Iprobe-Isat) did you plot it with respect to Vbias-Vfloating or just Vbias?
Apologies, but I don’t have any of this data. The post is dated 2015 because that’s when I converted my website to a new platform, but this work was done while I was a teaching assistant in the plasma lab at UCLA right around 2008. While it was a lot of fun to write this up, I haven’t worked in that lab, or with Langmuir probes, in many years. The plots are always plotted against V_bias directly.
Anmol Taploo says
I didnt round off data, i first got data from that excel file then based on saturation current, I substracted the given saturated current and took natural log to plot with Vbias. Then took slope in -22 to -17V range.
Yerbolat Ussenov says
Thanks for the nice explanation. However, I have question about the plasma potential.
If I’m not mistaken the plasma potential in gas discharge plasma is usually positive, because the plasma is positive respect to the wall. While the floating potential is negative due to the high mobility of electrons. So, you indicate that plasma potential is negative. Is it related to specific characteristics of your plasma source? Also, the positively biased part of you IV curve is very short and do you have any problem by collecting the electronica current data after increasing the applied voltage?
Thanks for the great question! This plasma was produced with an electron beam, so it’s reasonable to have a negative plasma potential. The measured current gets large as the bias moves more positive, and you see that it does not roll over like a flat ion saturation limit. The probe sheath likely expands as the bias gets too high, meaning that the measurements are less useful.
In order to extract the plasma potential, why do you take an intersection of linear fits ? For me, using the zero of the second derivative to find the inflection point would be more logical.
Great observation, and you are correct that you can identify the potential in that manner. If you had a nice digitizer and clean data set, then your method might be easier. If, however, you collected the data with an XY Recorder (https://davidpace.com/old-school-the-xy-recorder/) or had noisy data, then the graphical method is pretty good. Derivative calculations are prone to high variance unless you apply pre-processing such as smoothing or other techniques. The logical choice is whichever method works for your particular case.
Thank you for your feedback!
Abdul Qayyum says
Thanks, your explanation of I-V curve and extraction of plasma parameters helped me a lot. I have a question, for electron density calculation you used mass of the Ar ion. Why not the mass of electron?
That’s a great question! It is correctly the ion mass that matters for that calculation, but I don’t have an elegant explanation. Basically, it’s a force balance issue of the ion in an electric field generated by the electron density. As the ion mass increases, so too does the electron density required to draw it as a saturation current. Hope that is helpful,
Can you please tell me factors to be taken into account while deciding the R value??
The resistance, R, should be chosen based on your measurement setup. The value of that resistance will determine the range of voltage that you measure while performing the sweep. For example, your oscilloscope or other data recording device (even a voltmeter) will have some limited measurement range. The higher the resistance, the lower the current draw, which means that the voltage range will be lower as well. Generally, it’s better to start with larger values of resistance, and then to decrease it if you find that you want/need larger signal levels.
Hope this is helpful, good luck 🙂
Thank you 🙂
First,your posts are really helpful.
Thank you so much.
I am a bachelior student in S.Korea and try to simulate Langmuir probe that i found in Plasma Py site.Through that website i found your site and your sample data.I just wonder from what did you extract that data?Is it from real tokamka experiment?or something else?I want to simulate with datas from really tokamak whatever from ITER or KSTAR.
Thank you so much
Thank you, I am glad this information is helpful. The data from that sample was acquired from the undergraduate plasma laboratory course at UCLA (PHYS 180E). There is a photograph of the plasma device here, https://davidpace.com/ucla-physics-featured-in-physics-today/
In order to get data from a tokamak, you would have to contact the appropriate research team (typically a university) and ask for some. You might find a suitable group close to you by searching online.
Sh Hong says
Thank you so much for your answer.
I have one more question.
Your simulation is based on planar probe and graph should be flat,non?
I am simulating three geometries but it does not work as i expected.
I might be misunderstanding your question, but the example here is an analysis of measurements, it is not a simulation. The probe was a planar probe.
Sh hong says
Thank you so much for your answer.
I am studying this probe methode.
I want to try simulation.
I tried using plasmapy but with your data,it is not accurate.
And i am trying the simulation of three geometries.
It would be awesome,if you are willing to share the simulation codes of this post.
It looks perfect through my eyes,so i would ask.
My adress is email@example.com
My apologies, but I do not have this code. It was something I wrote while working in that plasma laboratory class. This is analysis, however, not a simulation code. This means that the code is really a representation of the equations shown here. The PlasmaPy version is probably pretty good, so you may want to double check your input. It’s pretty easy to corrupt a data set, e.g., maybe it was read incorrectly or formatted differently from whatever PlasmaPy expects. Best of luck to you, I hope you figure out the problem!
himanshu tyagi says
in our experimental RF plasma source, we are unable to get the knee point in electron region of IV curve. We find multiple slopes which cause issue in detection of floating potential.
Can you suggest some method in this case
This can be tricky, because there is no way to know whether your setup has a problem. Since your plasma is generated through an RF source, that can contribute noise to the signal, but it could also generate a non-Maxwellian distribution. You would have to find some literature about Langmuir probe measurements in RF plasmas and then determine what is relevant to your setup. For example, you might find something useful in Chen, Plasma Sources Sci. Technol. 18, 035012 (2009). Best of luck!
Hi, I have made one single langmuir probe and grabbing the values from ch2 and ch2 of my expeyes junior data acquisition card..but u have some problems in it..can you help me out..if yes please please let me know yours email ID.
It is very difficult to help someone with a hardware setup through email. I don’t have any experience with EXPEYES. Thank you for mentioning that, however, because I looked it up and that is a neat system.