RGAs and Leak detectors

[Note that standard Ion Implanters are just overgrown RGAs!]

RGAs or Residual Gas Analyzers are also known as Mass Spectrum Analyzers. These can sometimes be upgraded to also include energy analysis – in which case they are known as Mass and Energy analyzers.) While the portion of the RGA that is used to perform the mass analysis differs from tool to tool there are many similarities between all RGAs. The basic system looks like:

a)b)

Figure 1: The basic layout of a Residual Gas Analyzer. (a) Schematic diagram identifying the major components of an RGA and showing typical ion motion. (b) Picture of an old, half disassembled RGA showing the ionizer region (left) and the quadrupole rods in the mass resolving region. The detector region has been removed from the right hand side end.

We will go through the ionization part of the RGA first. The plasma produces a large variety of chemical species many of which are neutrals. To examine these neutrals we must be able to apply some well characterized forces to them. The simplest thing to do is to ionize the molecules and then use electromagnetic forces to push them around in the mass resolving section. [NOTE: if you turn off the ionization region than the only thing that you can look at is ions coming out of the plasma. This a useful variant to standard RGAs.] To ionize the neutrals we use fast electrons. These electrons are created by first heating a filament or wire that is known to give off large numbers of electrons at high temperatures. This is known as Thermionic emission. Common materials used for this is Tungsten, Thoriated-Tungsten (1-5% Thorium), or Lanthanum-hexa-Boride (LaB6). Thoriated-Tungsten is perhaps the most common of these materials. This is because of it very high emission coefficient, higher than plain Tungsten, and it does not suffer the same material poisoning problems found with LaB6 emitters. [Note that LaB6 emitters are common in other applications.] To heat it, a current is drawn through the filament. Often the bias is on the order of 20 V and perhaps an amp of current but this depends on the size and type of the filament. The typical temperature of the filament – at typical operating conditions – is 1600 to 1800 ° C. This is in color a very bright red to a white. When the filament reaches this temperature the electrons in the filament have sufficient energy to overcome the binding energies – and will leave the surface with a fraction of an eV of energy. (1 eV is energy gained by dropping an item charged to one electron charge through a voltage of of one volt.) Because we wish to ionize the gas in the ionization region, and this often takes 10s of eV in energy, we need to accelerate the emitted electrons. We do this with a bias voltage that is typically in the range of 70 to 120 V. By using a grounding cage around the ionization region, we assure that most of the electrons enter the region with energy near the bias voltage (times ‘e’). This bias voltage is chosen because the maximum in the ionization cross section for most gases is near 100 eV. It is very important to note that the maximum varies both in magnitude and peak location. Thus we will get different numbers of ions from different gases – see Figure below.

If we can determine the number of ions produced per neutral gas molecule we can determine the partial pressure of each of the species in the plasma. If our electrons suffer a number of collisions as they cross the ionization region than we will have a large range of energies at which the ionization collisions might take place. This would make it very difficult to determine the ratios of the densities of different species. Thus we need to make sure that the pressure in the ionization region is low enough that the mean-free path for ALL collisions is long compared to the total distance traveled by the electrons. Often at process conditions the background pressure it too high for this requirement – thus we often have a pin hole entrance aperture and use a high vacuum pump to drop the the RGA to pressures near 10 µTorr. It is important to note that the HV pump is most often a Turbo-molecular pump because of those pumps pump most gases at the same rate and they do not add significant amounts of contaminates. [Ion implanters do not have this limitation. There we simply want to create as many ions of a specific species as possible. We are not trying to preserve density ratios.]

IN ADDITION TO IONIZING THE GAS, THIS SECTION ALSO BREAKS UP THE GAS INTO DAUGHTER SPECIES. THIS IS KNOWN AS ‘CRACKING’ AND WE WILL TALK ABOUT IT DURING THE NEXT CLASS PERIOD.

We have now created ions which exist at densities that small compared to the densities of the original gases. This ratio is of course set by the ionization cross section of the gas and the cracking pattern of the gas. (Some species that are stable as neutrals are not stable as ions and immediately fall apart.)

We now need to get the ions to our mass analysis region. To do this we accelerate the ion to approximately 20 eV using electrostatic ion ‘optics’. [For implanters this is usually ~20 keV. This is because we can get more ions/area through the system at higher energies.] In its simplest form this amounts to a negatively biased plate with a hole in the center. Note that the rest of the RGA sits at this same potential… Thus when the ions enter the mass resolver and detector we do not disturb the orbit of the ions with this electrostatic field.

At this point we enter the mass resolution system. In reality this is a velocity resolution system – we use the fact that mass, energy and velocity are intertwined and we have set the energy using the ion optics. While there are numerous types of velocity analyzers, two are most commonly used. The first uses a static magnetic field or magnetic sector – we will go through the math on this one in a moment – while the second uses a quadrupole electric field containing both dc and rf parts. The later of these two systems is the most common in good RGAs. However, the analysis of how they operate is beyond the scope of the class. The Magnetic sector systems are not typically used in good RGAs BUT they are the mainstay of ion implantation mass resolution systems. We will go through the math on how these systems work here.

The equation of motion for a single particle of charge q and mass m is given by the Lorentz force law:.

Cases 2:

Thus we find

To solve this set of equations, we must separate the components of the velocity. This is simple to do by differentiating the equations again and substituting to give.

These second order equations are of course are easily solved as

Now taking into account the original coupled first order equations we find

Integrating a second time we find,

It is easy to see that is the radius of a circular orbit around the magnetic field line; it is known as the Larmor radius or the cyclotron radius. Further we note that the positively charged particles orbit in a left-hand orbit while the negatively charged particles orbit in a right-hand orbit.

Now for our system the velocity is set by energy we have given the ion with the electrostatic accelerator. Thus we find that the particle travels in a circular orbit with a radius of

Thus we can change the radius by either adjusting the magnetic field strength or by adjusting the accelerating voltage. Often some combination is used. By setting up out analyzer as thus:

We can eliminate all ions which do not match the required Larmour radius.

At this point we need to detect the ions that have made it all the way through the system. The simplest detection scheme is to use a Faraday cup. This amount to a cup that simply collects all of the ions, and we measure the associated current.

The next thing that one might use is an electron multiplier – note that these are also used on photomultiplier tubes to measure photons. There are two main types the chain and the channeltron. Both make use of materials which produce significant quantities of secondary electrons when struck by either energetic ions or energetic electrons. These items greatly increase our currents – allowing use to measure single ion events. (Electron multiplier require high vacuum for a number reasons…

Add drawings of chain and Channeltron tubes.

We can change our detection system to include electrostatic retarding optics. By varying the potential on our retarding optics, we can stop some of the ions from entering the detection area. By measuring the current as a function of the retarding potential we can determine the energy distribution of the ions entering the detector – thereby allowing us to also measure the energy.

Cracking patterns

Interpretation of RGA data is fairly complex. First, the same type of atom can have a variety of masses. As an example most of carbon is of the type C12. This means that the carbon is of mass 12 AMU – atomic units. C12 has 6 protons and 6 neutrons in the nucleus or core of the atom. However approximately 1% of all carbon on earth is C13. (Note that other places in the universe this ratio might be very different!). Because of these different atomic masses when ever we have a carbon containing molecule we will have the spectrum which assumes C12 and a second set, at 1% of the height, which assumes C13. Note that C is common in many etch gases, CF4, CHF3, C2F6 and all of the other halogens. For simplicity let us assume that we have a little H2O in our chamber. We will have a pattern something like what is given below. Note that slightly less than 1% of the main peak (18 AMU) is also found at 20 AMU. We also have a large number of total peaks – 9! And this is for a single molecular species – a simple one at that.

At this point we might sit and think about what the RGA does to produce ions – it bombards neutrals with high energy (70 – 120 eV) electrons. While the purpose of these electrons is to ionize the gas, other collision types are possible.

Elastic collisions take < 0.01 eV of energy from the electron but leave the neutrals pretty much intact.

Inelastic collisions are not so nice. We have

Excitation

H2O +e -> H2O* + e

Ionization

H2O +e -> H2O+ + 2e

H2O +e -> H+HO+ + 2e <= use in next graph @ 5% of peak H2O+

H2O +e -> H+ + HO + 2e

H2O +e -> H+ + H + O + 2e

H2O +e -> H + H + O+ + 2e <= use in next graph @ 20% of peak H2O+

H2O +e -> H2 + O+ + 2e

H2O +e -> H2+ + O + 2e <= use in next graph @ 1% of peak H2O+

How each of these species are formed depends on the cross section of each process. (Note that we have ignored the chemical breaking collisions that do not result in ions. Ions can still be produced from those daughter species by a second electron impact. This would be of the form.

H2O +e -> H2 + O + e

O + e -> O+ + 2e

If we where to look at how this cracking affects the result – picking only three of the possibilities – we get:

Now let us continue this game. Some of these species can become doubly ionized. Let us assume that 1% of H2O+ is H2O++. At this point we get

This is much more complex and remember that we only have a single species present in the chamber!

Fortunately life is not this complex. Usually the cracking patterns are such that only a few lines are present for a given species and the ratio of the lines – while RGA system dependent – are relatively uniform from RGA to RGA. Some of the more common spectra are given in Appendix E of J.O. (H2O is in E.3 and is not as complex as we might have guessed.)

Quantitative Analysis of RGA data

True high quality quantitative analysis of RGA data is very difficult. To do this one must produce cracking patterns for each and every gas that might be present in the chamber. This means that if we are using CF4 and producing CF3 in the plasma, we really need to get the cracking pattern for CF4 and for CF3. (CF3 is a radical – thus one can not have a ‘bottle’ of the stuff sitting on the desk. That said, this means that we need to produce CF3 in a controlled manner and introduce this to the RGA for analysis of the cracking pattern. This is tough!) By measuring all of the cracking patterns for all of the species, we can then begin to try to untangle the RGA spectrum and thus get a high quality quantitative measure of the species present in the chamber… Such an experiment might take years. On the other hand one can get an approximate measure of the species concentrations by assuming that the maximum line for each species has approximately the same signal strength / molecule. This implies that if we were to have a line strength for AMU 18 (H2O) of ‘1’ and a line strength of 0.5 for AMU 28 (CO) then the density of H2O is twice that of CO. This is at best a ballpark estimate. If one then knows the pressure in the chamber, Pchamber, one can assume that the partial pressure of the H20, PH2O ~ 2/3 Pchamber and the partial pressure of the CO, PCO ~ 1/3 Pchamber. This is at best an approximate estimate – remember the ionization cross sections can be very different as can the be the transmission factor for each species through the mass analyzer. However, this will give ballpark figures for the partial pressures. These figures can be improved by examining the literature for the cross sections for ionization for each species. Then all one must do is assume that each species travels through the analyzer at the same rate and is measured by the detector at the same rate. Such errors might be only 50%. This process become more difficult if the lines over lap…