Wednesday, June 20, 2018

XPS Depth Profiling: Thin Ir Films

In addition to angle-resolved XPS (ARXPS), the chemistry of surfaces and interfaces can be studied using depth profiling. In this case an Ar+ ion beam is used to sputter the sample surface in between alternating XPS studies. To accomplish this the sputtering conditions must be optimized for both the materials and depth ranges involved in the study.

As an example, higher beam energies yield higher sputtering yields-- the number of atoms sputtered per incident ion-- but a high yield will just cause one to punch through a very thin layer without getting any data from that layer.  Thus while great for speed, high sputtering yield is bad for very fine depth resolution of thin layers.

Also, geometric relationships matter. If the sample is not rotated then it will crater asymmetrically towards the ion-gun because the ion density is larger closer to the ion gun. Azimuthal rotation, called Zalar rotation, will reduce this cratering-- but at the cost of lower yield. Another geometric concern is the angle of ion incidence. When this angle is lower and more grazing, there is less transverse momentum transfer to the sample and then the sputtering yield is much lower. Lower incidence angles are also more useful for rougher samples.

The size of the sputtering crater relative to the analytical area is also important. The sputtering crater  needs to be larger than the analysis area and the analysis area needs to be centered on the sputtering crater. If this is not the case, then the analysis region will include the crater wall and will include information from different depths.

To accomplish this, great effort must be made to align the ion-gun at the working stage height, calibrating the ion-gun rastering, and optimizing the ion-gun condenser and objective lens settings to provide a well defined and focused beam of specified current.

In this study, a 4 nm iridium film was sputtered through a 10 mm diameter mask directly onto a polished 304 stainless steel XPS sample stub. A 1 keV Ar+ ion beam was rastered across the sample to produce an approximately 13 mm diameter region of sputtering. The intention is to guarantee the complete sputtering of the Ir-deposited region with minimum decrease in current density. The Zalar rotation was set at a few RPM and at an incidence angle of 45 degrees. The ion current was estimated to be 1 µA looking at the current deposited onto graphite which has minimal secondary electron yield, allowing the stage ammeter to more closely reflect the ion current.

The sample was sputtered for 1 minute and then data taken in the Ir 4f, Fe 2p3/2 and O 1s spectral regions using a constant 71.55 eV pass energy. Baselines were fit to the peaks and using peak heights and corresponding sensitivity factors that reflect photo-electron cross sections and the analyzer transfer function, the spectral regions were quantified to reflect atomic percentages.

The NIST Standard Reference Database 71 can be used to determine the inelastic mean free path (IMFP) of the Ir 4f photo-electrons using the Gries predictive model. The Ir film thickness of ~ 4 nm is just ~3X the IMFP, so the entire depth of the film is expected to be detected by XPS from the onset of the depth profiling. Such a thin film is effectively "transparent" in XPS, and this is supported by the continuous and nearly linear decline in the Ir concentration through the depth profile. The Fe concentration similarly increases in a nearly linear fashion.

Half the iridium is sputtered down in about 70 minutes, which gives a yield of 0.3 Å/minute. It is so low because of the decreased current density to sputter this 10mm disk of deposited iridium. But because of the care in conditioning the ion beam we have a linear sputtering rate through more than 140 minutes of sputtering through just 40 Å of material.

Wednesday, June 11, 2014

Ghost Peaks

In XPS, unlike Auger spectroscopy where the kinetic energy of electrons is presented, the binding energy of electrons is presented. In this case, the kinetic energy of the photoelectrons is measured by the analyzer and then converted to binding energy by referencing to the energy of the X-ray excitation source.  The relation is:

BE = hν - KE - ϕ

where ϕ is the work function of the XPS analyzer. In practice ϕ simply becomes a calibration factor to place the peaks of standard materials at the proper binding energy. The Mg Kα  line used is at 1253.6 eV and the Al Kα  line is at 1486.6 eV.

It's possible, because of the structure of the X-ray source, for electrons and/or X-rays to strike some part of the source interior other than the intended anode. When this occurs, X-rays are produced which will generate XPS core level peaks as well as Auger peaks. However, since the binding energy is referenced to the X-ray energy of the chosen, these peaks will no be at the right positions. These are called ghost peaks.

For a given anode, these ghost peaks occur at a fixed binding energy interval above and below each peak. In this study, graphite was flash coated on a 304 SS stub and sputter cleaned. Some residual O 1s signal was present because of water physisorbed in the graphite film.

In the first graph, 233.0 eV above the C 1s peak, just below the O 1s peak, is an unidentified peak. It's too intense to the a satellite to the O 1s peak, and remains upon satellite deconvolution. Possible assignments include V 2p and Re 4p-- which are ridiculous. This is actually a ghost peak from Mg Kα radiation being generated within the X-ray source when set on Al Kα. If we look at the structure of the X-ray source, what is happening is that some small fraction of the electrons generated by the X-ray source filament are making it over the bridge separating the Al and Mg anodes and making Mg Kα X-rays. Comparing the area of this peak to the main C 1s peak, it amounts to about 1.5% of the total X-ray flux. As such the Mg Kα ghosts when using the Al source are an issue only with the stronger detected lines. If one encounters a peak that one can not assign when using the Al anode, subtract 233.0 eV and see if there is a significant peak.

The middle picture shows a Cu Lα ghost peak for the Al anode. It is 556.9 eV above the C 1s peak generated by Al Kα. This peak only reflects about 1% or less of the total X-ray flux, and is an important diagnostic of the health of the anode. The Al anode material is actually attached to a copper block that is used for heat transfer to the water cooling. Over time the Al anode material can pit or erode by essentially sublimating if there is not sufficient water cooling power. The fact that we are seeing 1:100 Cu La ghost intensity indicates that the anode is in good condition. If one encounters a peak that one can not assign when using the Al anode, subtract 556.9 eV and see if there is a significant peak. In this case possible assignments would have included Co, Gd and Cs Auger peaks-- all ridiculous.

The last picture shows an O Kα ghost 961.7 eV. This is due to X-rays being generated by oxidized portions of the X-ray source interior. This amounts to about 2.5% of the total X-ray flux, and is another measure of the health of the anode. Over time, if the XPS chamber is too high in pressure, the anode can become oxidized, as can other portions of the source subject to heating such as the cap that holds the window. Possible assignments would be Xe 3d or Ga 2p1/2-- all ridiculous. Again, if one finds a peak that one can not reasonably assign, look for ghost peaks. If one encounters a peak that one can not assign when using the Al anode, subtract 961.7 eV and see if there is a significant peak.

If peaks assign to bizarre elements that are not likely present, first look at X-ray satellites, and then ghost peaks.

The Mg anode has it's own potential set of ghost peaks. O Kα at + 728.7 eV, Cu Lα at + 323.9 eV, and Al Kα at -233.0 eV.

Tuesday, June 10, 2014

Thermal Desorption

Metal surfaces are prone to physisorption of molecules with dipole moments. In the case of physisorption, the electronic structure of both the adsorbed molecule and the surface are negligibly perturbed. The mechanism of absorption is between the dipole moment of the molecule and its image in the metallic surface.

Water is the most classic example as all samples are prone to physisorption of atmospheric water. Water has a dipole moment of 1.85 D. One challenge of studying the physisorption and thermally assisted desorption of water from metal surfaces is that many metals have a native oxide barrier. As such the O 1s signal of the native oxide must be separated from that of the physisorbed water, and the presence of water itself can result in surface oxidation.

In the present study chloroform, CHCl3, which has a significant dipole moment of 1.15 D was pipetted onto a UHV clean 304 SS XPS stub. As seen above in the unsputtered XPS data taken with Al Kα  X-rays, there is a strong presence of Cl 2p and Cl 2s XPS lines showing the presence of the physisorbed CHCl3.

Instead of sputter cleaning the sample, the sample was pulled into a side chamber and heated at 350 C at about 4 x 10-5 torr. After 30 minutes the C 1s peak narrows and a higher binding energy satellite appears while the Cl 2s and 2p peaks decrease. After heating for 60 minutes there is just a trace of Cl 2p and the Ar 2p peak from previous sputtering appears. Unsputtered UHV cleaned 304 SS-- water and detergent, DI water, acetone, and EtOH-- is known to have a significant amount of C 1s contamination.  The presence of this satellite around ~ 296 eV is a little high for a C 1s chemical shift, unless it is due to charging, but it is consistent with an emergent K 2p line.

The purpose of this study is to show that thermal desorption is an effective and alternative means of cleaning samples, and can be used to selectively remove physisorbed material while leaving native oxides.  While heating at much higher temperatures can be used to remove even native oxides, the present oven does not go about about 350 C.

Friday, April 25, 2014

Ion Gun Raster Calibration

The sputtering yield is a function of current density, so rastering only the area required for one's application is desirable to optimize sputtering yield. This is particularly important when depth profiling to minimize the time required to do a depth profile.

To calibrate the rastering area, a thin layer of Au was sputtered onto a 304 stainless steel sample puck. Zalar rotation was turned on to produce a circular sputtering pattern since the sputtered area is actually a rectangle, the diagonal of which determines the diameter of the sputtered circle (presuming the gun is centered on the center of the puck). When the Au layer is sputtered through, a change in color is evident.

In this image the X & Y rastering was set to 15 mm and a sputtered circle of approximately 12 mm in diameter was produced. This experiment was repeated for 10 mm and 20 mm rastering sizes.

Fitting the estimated diameters vs the rastering size, the sputtered disk is approximately 85% of the set rastering size. Inverting the problem, one should set the rastering size to approximately 1.2X the the diameter of the sputtering region desired.

This was performed at 3 keV ion beam energy, but should apply to all energies. It was also performed at a stage height of 15.

Wednesday, August 14, 2013

Mounting Powder Samples for XPS

There are several methods for mounting powder samples for XPS. One is to press the powder into a very thin metal foil that is malleable enough to allow the powder particles to impregnate the foil. Indium is the most common choice, and in this case the sample powder is rolled or pressed into the indium foil and that powder-impregnated foil is used as the XPS sample. Since no adhesive is used, this is a truly "ultra high vacuum" method of sample mounting. One problem with this method is that indium, being a fairly high-Z material, can provide significant spectral contamination from its many XPS and Auger peaks.

To remedy the complication of pressing samples into metal foils, as well as the corresponding spectral contamination, we have successfully used double-stick carbon tape used in SEM sample preparation. The mounting procedure is shown below:

Step 1: Place a disk of double-stick carbon tape on XPS stub. It should be centered on the stub so that the ion-gun rastering can be easily reduced to facilitate sputtering just the sample, thus increasing the ion-density and thus the sputtering yield.

Since the carbon-tape disks will be introduced into an UHV environment, it is best to keep the disks in a dry environment such as a desiccator. The disk pictured is roughly 1 cm in diameter. Rub and press the adhesive disk onto the XPS stub. If the XPS is coated, the adhesive disk might not stick as it may remove the stub coating. This is acceptable. Let the adhesive disk remove the coating-- just replace the adhesive disk where the film has been removed from the XPS stub.

Step 2: Using a spatula, place the powder on the exposed adhesive disk. Again, since XPS requires a UHV environment, the powder needs to be appropriate for that environment. If the sample is inherently porous, it should be baked out under vacuum to remove as much physisorbed water and solvent as possible. After baking under vacuum such powders should be stored in vacuum or in a desiccator. As this is a UHV technique, manipulate the powder only with UHV clean spatulas.

Some materials simply have very poor vacuum characteristics. Use the ion-gauge and residual gas analyzer look for signs of problems before introducing the sample into the main XPS analysis chamber. Is the base pressure too high? Has the sample visibly changed appearance? If a problem is suspected, the RGA will give a fingerprint of the intro-chamber vacuum environment. If there is a huge background of water or organics-- do not introduce the sample into the XPS.

Step 3: A pile of powder will not make it into the the XPS safely.  It should be noted that powders migrating through the XPS vacuum system is potentially very destructive to the instrument. Migrating powder can land on valve sealing surfaces causing them to leak and damage them so that they can never seal. Migrating powders can also destroy the turbo-molecular pump and the mechanical pump that backs it.

To properly prepare the powder sample, use a spatula to smear the powder across the surface of the adhesive disk of carbon tape. Distribute the powder as evenly as possible and uniformly cover the carbon adhesive. A failure to do so will result in significant C 1s, and to a smaller extent O 1s, XPS background from the tape. Note that the stub shown in the third image can not and should not be placed in the XPS intro-chamber. The unattached powder can migrate into the vacuum system from the laminar flow during pump-down as well as through the vibration of pneumatic valves opening and closing. By "damage" I don't mean some down time and an interruption of the work flow-- I mean that plus as much as $20,000 of damage.

To get the stub ready for the XPS, tap the stub very hard on a rigid surface. Do this until there is no evidence of any powder coming off the surface. Tapping on a white weight paper can help. While it may seem that a good deal of powder has fallen off-- that is acceptable. The XPS is a surface technique that looks at a few nano-meters of the sample surface. A fine dusting of material, seemingly insignificant to the naked eye, is sufficient. If your sample is multiphase and one component is a fine powder while other is larger chunks-- then this can lead to some sampling bias, so consider this in sample preparation.

Thursday, May 3, 2012

Contamination of X-ray Source Window

This is an SEM image of the window of the XPS X-ray source. It consists of a thin Al window on a BeCu alloy mesh and frame to support it. Al is transparent to the Al Kα and Mg Kα X-rays and provides a conducting surface to trap any electrons from the source. Such electrons could hit the sample and generate XPS core level electrons as well as Auger electrons and other low energy electrons. In the top right of the image one can see damage to the window. While it is in service, it is exposed to IR (heat) radiation, UV, and X-ray radiation up to 15 keV. The aging of the window can be see in a  previous post which addresses window replacement.

EDS was taken on different portions of the window. Spot #1 is take on the window away from the BeCu frame. The Al Kα line at 1486 eV is clearly dominant and represents the window material. What is striking is the amount of spectral contamination.

Au Mα at 2120 eV and Ag Lα  at 2983 eV are clearly visible just above the Al Kα  peak. These peaks arise from sputtering Au and Ag coated pucks used in both routine XPS analysis as well as calibration. The Ag Lα line overlaps with the Ar Kα line, so it's possible that some small portion this peak also reflects Ar+ embedded in the window material. The Fe Kα line is also visible at 6398 eV. This is probably the result of contamination of the XPS X-ray source window from sputtering 304 SS (stainless steel) sample pucks. Also present is a hint of W, which would likely be due to the sublimation of the W filament onto the back of the window-- at 30 keV the electron probe is going right through this thin window, so material on the back of the window is being excited.

Spot #2 shows EDS on the grid region. As such, the spectral information should reflect both the window and grid composition. The spectrum is similar except for the presence of the Cu Kα and
Kβ peaks at 8040 eV and 8903 eV respectively, and the Cu Lα peak at 930 eV.

There is one subtle difference regarding the signal from the window: the W Lα peak seen at 8394 eV at Spot #1 is missing at Spot #2. This supports the hypothesis that at Spot #1 the e-beam is exciting W sublimated from the source filament onto the back of the window since the e-beam does not pass through to the back of the BeCu grid at Spot #2 to excite the W deposited there. A small trace of Si Kα is visible in the absence of W Mα which is probably due to contamination of the window by sputtering samples on Si wafers and glass.

The point of this note is to illustrate the importance of raising the X-ray source when sputtering-- especially for extended periods of time. As far as X-rays go, Al Kα and Mg Kα used in XPS are fairly low energy and the high Z elements found on the X-ray window-- Au, Ag, Fe, W-- are all capable of attenuating the excitation source. Additionally, while the Al Kα and Mg Kα lines are too low in energy to excite the higher energy higher Z lines, the XPS X-ray source does operate at 15 keV so there are higher energy continuum X-rays available to excite these lines.  The presence of Au Mα, Ag Lα or Fe Kα in the excitation source will lead to spectral ghost peaks.

Please look at this related post that addresses raising and lowering the X-ray source.

Wednesday, February 8, 2012

X-ray Source and Ion Gun Geometric Considerations

The distance between the X-ray source and the sample can be changed using the linear motion feed through indicated in the first image in this application note. The closer the sample the higher the X-ray flux on the sample surface. This will lead to higher counts and as a consequence a higher trace sensitivity. Other smaller spectral features such as plasmons, escape loss peaks, spectral satellites, ghost peaks, etc. will be easier to see.

On the other hand, the closer the X-ray source is other problems can occur. One is the heat deposited into the sample. While a larger solid angle of X-rays is projected onto the surface, so is a larger solid angle of IR and UV. As a consequence there is the possibility for sample heating. The source should thus be pulled back in the case of very delicate samples that involve physisorbed species of interest as they might desorb upon slight heating.

The second image shows another problem. The X-ray source can be so close to the sample that its form factor can actually block photoelectrons from being collected by the XPS analyzer electron optic. While it would be impossible to get the source so close that no signal were detectable, it is possible for the signal from the side of the sample stub opposite to the analyzer to be blocked by the X-ray source. If samples are not mounted near the center of the stub this can lead to pathological oscillations in XPS depth profiling when using Zalar rotation. With the rotation of the stub a radially displaced sample can move in and out of the blind spot of the analyzer.

The final picture shows an additional problem. The X-ray source can be so close to the sample that it blocks part of the ion-gun beam. This has several consequences. One is that it can then be difficult to properly sputter the sample surface. The other is that portions of the X-ray source itself can be sputtered! The X-ray source has openings to allow gasses evolved from the anode and filaments to be removed, and the interior portions of the X-ray source can be hit by the ion beam. This can't be good for the long-term health of the X-ray source, and when doing XPS while sputtering can result in a shut-down of the X-ray source.

All of the images above illustrate another problem-- when the X-ray source is very close to the sample during ion sputtering, the X-ray window will gradually become coated with sputtering products. This can lead to a variety of complexities. One, the material sputtered onto the X-ray source window can fluoresce unusual wavelengths of X-rays producing X-ray artifacts. While a good XPS operator would be accustomed to looking for Al-Mg anode cross talk and spectral contamination from W, Fe and Cu-- if one were sputtering GaAs one would probably not think to look for spectral features from Ga Lα and As Lα X-rays produced in the X-ray window. Another obvious consequence is that the X-ray source for XPS is quite low energy- 1.2-1.4 keV-- and an adlayer of material on the X-ray window will greatly attenuate the X-ray flux available for XPS.

If one is working with very small X-ray source to sample distances to increase counts, then it is imperative to back away the X-ray source during sputtering.

New Turbo & Intro Pumpdown Procedure

We have a new Pfeiffer HiPace 80 turbopump on the XPS intro-chamber. To prevent future turbo failures we have slightly modified the pump down procedure. The pumping line between the turbo and its backing mechanical pump is diverted to the intro-chamber to allow for rough pumping of the chamber. If this chamber roughing valve is opened a little too much too quickly it can cause significant braking of the turbo due to increased gas load. While this gas load shouldn't be enough to damage the turbo, it is loading the turbo opposite to normal gas load-- through the turbo backing port.

The first step in the new pump down procedure is to close the intro-chamber door. In the first photo it is shown unlatched. Make sure the N2 vent gas pressure is only 1-2 PSI. If one latches the chamber while venting to very high pressures of N2, the window on the intro-chamber could rupture. Latch the door. The latch is circled in yellow.

The next step is to close the valve between the mechanical pump and the backing port of the turbo pump. This valve is circled in yellow in the second picture. Turn it in the direction of the yellow arrow to close it. This will allow high pressure gas in the intro chamber to follow the path of the blue arrows into the mechanical pump which is off to the left of the image.

If this valve were not closed then high pressure gas from the intro-chamber could follow the red arrows and enter the turbo pump through its backing port. This is problematic as it will shock the bearings of the turbo. It will also allow particles and fibers from powder and fiber samples to enter the turbo where they can migrate into the bearings and shorten their lifetime. Since the turbo is not under any gas load it can continue to run without being turned off. Do not turn off the turbo pump controller!

The next step is to rough pump the intro chamber. Now that the tubro is isolated from its backing pump, the roughing valve to the intro-chamber can be opened. This valve directly connects the intro-chamber to the mechanical pump. It is circled in yellow in the fourth image. Turn it in the direction of the yellow arrow to open it. As the knob of this valve moves upwards, it will open the switch circled in green. This switch turns off the flow of venting N2 into the intro-chamber. However, V2, the venting valve will still be shown to be open on the automatic valve controller (shown below). Open the roughing valve all the way, allowing the gas in the intro to follow the blue arrows to the mechanical pump. The intro chamber pressure is read by convection "A" shown in the image. A base pressure of low 2-3 x 10-2 torr should be reached in a few minutes.

When the intro chamber reaches its base pressure using the mechanical pump, it's time to switch to turbo pumping. Go to the side of the XPS console and open the valve at the turbo roughing port. This is the black-handled valve shown in the second image above. Turn it in the opposite direction as the yellow arrow to open it.

Then go to the automatic valve control panel shown at the end of this application note. The red arrow indicates the V2 venting valve as being open-- but this was bypassed when we opened intro chamber roughing valve as discussed above.

The "pump intro" button is circled in green. Press this and then close the intro chamber roughing valve shown in the third image. Turn it in the opposite direction as the yellow arrow to close it. Now the intro-chamber is being turbo-pumped. V1 and V2 should be red and V3 should be green showing that the intro chamber is open to the turbo pump.

The red LED's circled in blue will start to light as a thermocouple gauge below the table top measures a better and better vacuum. When all are lit the pressure interlock for introducing samples is met-- but it is recommended to wait until the base pressure is in the low to mid 10-6 torr range before pressing "intro sample" and attempting to introduce/remove a sample. The intro chamber ion gauge will have to be turned on to measure the intro chamber ultimate base pressure using the turbo.

In summary:
  • Close the turbo backing valve.
  • Open intro roughing valve.
  • Reach intro roughing base pressure.
  • Open the turbo backing valve.
  • Press "pump intro" on automatic valve controller.
  • Close intro roughing valve.
  • Wait to achieve base pressure using turbo.
Add Image

Monday, February 6, 2012

Changing Anodes: New Software

The new RBD AugerScan software has a slightly different method for selecting anodes.

As before, select Hardware Properties on the low bar of icons as circled in red. The pop-up shown will appear. There is a new section that allows one to select filament #1 or filament #2. These options are hardware determined. Filament #1 is Mg Kα and filament #2 is Al Kα. The filament selection is circled in blue. Below that, also circled in blue is the assignment of that filament. One can select Mg or Al and assign that photon energy, work function and X-ray generator power to the filament selected above.

Make sure these photon energies are correct. Mg Kα is 1253.6 eV and Al Kα is 1486.6 eV. The work function depends upon the current calibration which is a function of tweeks on the analyzer boards. 400 W is full power for the X-ray source.

This is a more flexible version of the software-- but it allows one to take data with the Mg anode while applying the Al photon energy in calculating the binding energies and vice versa!

Thursday, February 2, 2012

Bulk and Surface Plasmons

Peaks to the low binding energy side of an XPS peak are potentially XPS satellite peaks. These have been discussed in a previous post. These peaks contain redundant spectroscopic information as they come from the same core levels but due to different X-ray lines because the source is not monochromatic. These peaks are always the same number of eV in binding energy away from the primary XPS line since there is a fixed discrete relationship between the photon energies in the non-monochromated X-ray excitation source. The AugerScan and CASA packages have deconvolutional filters for removing these X-ray satellite peaks.

While XPS satellite peaks occur whenever an XPS core level peak is visible, in cases where a sample surface is exceedingly clean, special loss structures can be see on the high binding energy side of the XPS peaks. These are due to the photo-electrons leaving the surface exciting plasmons-- collective excitations of the electron gas. There are plasmons associated with bulk and surface electronic states, and depending upon the resolution of the XPS energy analyzer these states may or not be resolved.

This image shows the loss structures associated with both surface and bulk plasmons from a very clean silver surface. Bulk and surface plasmons are unresolved. Note that the splitting of these two plasmon peaks equals the doublet splitting of the 3d XPS peak. Each loss structure corresponds partly to bulk and surface plasmon losses of the Ag 3d5/2 and Ag 3d3/2 XPS peaks.

Plasmon peaks are more than just a curiosity. They can be used as a very sensitive measure of surface chemistry and knowledge of their potential existence can prevent confusing misidentification of plasmon peaks as photoelectron peaks. In general these loss structures on the high binding energy side of an XPS peak can form a broad loss peak that represents the aggregate of energy loss of the photoelectrons leaving the surface. In most materials this is not resolved as discrete surface and bulk plasmons, just a broad hump that is the loss peak.

Tuesday, December 13, 2011

Low Energy Ion Neutralization: RuCl3 Insulating Sample

In EDS charging of the sample is not an acute problem since X-rays are not charged and do not experience a shift in energy when being emitted from a sample at potential. In EDS sample charging becomes an issue when it is so extreme that the probe is deflected and one can not be certain of the current in the excitation area or the location of the excitation area itself. In such cases quantitation becomes impossible, but not the identification of X-ray lines.

With XPS we have the opposite problem. The excitation source, X-rays, is charge neutral, and has no problem exciting a charged sample. However, because of the high energy resolution of surface electron spectroscopies, the deceleration of the photo-electrons leaving a positively charged sample manifests as something like a chemical shift-- a shift towards higher binding energy. These charging shifts can be significant, equal to or larger than chemical shifts, and can complicate the identification of chemical shifts.

One approach in dealing with charging shifts is to look at the corresponding BE shift of C 1s which is typically adventitiously present on all samples. The shift of C 1s can then be used to calibrate the BE of other shifted peaks in the system. One problem is that carbon tape is often used to mount powder systems. As this C 1s signal comes from a conducting part of the sample, unless there is significant C 1s contamination of the insulating sample one will always see the C 1s signal at ~ 284 eV. Other methods are needed.

The FIG 5 CE ion gun has a mode that acts as a charge neutralizer. The column can be floated allowing for the production of very low energy ions. These low energy ions are not energetic enough to sputter the sample or generate secondary electrons (and thus making the sample charging worse), but are able to donate charge to the sample and neutralize the charging induced during an XPS experiment.

The figure above shows the XPS Cl 2p region of RuCl3 taken with Mg Kα. The blue curve shows the Cl 2p peak taken without any charge neutralization, and shows the XPS peak shifted by ~ +5 eV due to sample charging. The green curve shows the same data taken with a 10 eV neutralizing beam. For comparison, the red curve shows the same data taken with a 500 eV Ar+ ion beam which makes the charging worse, shifting the Cl 2p XPS peak by ~ +20 eV.

The bottom image shows the Ru 3d region of RuCl3 taken with Mg Kα. The data is shown again without the 10 eV neutralizer beam and with the 500 eV Ar+ beam which makes charging matters worse. The Ru 3d5/2 peak is shifted by ~ +5 eV just as was the Cl 2p peak. As above, the 10 eV neutralizer brings the Ru 3d5/2 peak for RuCl3 very close to it's tabulated value minimizing the effects of charging.

This example shows another useful application of the ion gun. Charging shifts only apply to insulating portions of the sample. The 500 eV ion beam shifts the Ru 3d doublet by about 20 eV, making the C1s peak from the carbon tape visible. This portion of the sample probed by the XPS did not shift because it is conducting.

Point of this application note: use the FIG 5 CE to neutralize charge using a very low energy neutralizing beam-- or use larger energies to larger separate the signal from conducting and insulating parts of the sample.

Thanks to Dr. Jim Zheng, FSU COE for the RuCl3 sample.

Tuesday, December 6, 2011

Zalar Rotation

The CMMP XPS has a custom-made Zalar rotation stage for depth profiling. This stage was designed and fabricated by Ian Winger of the FSU Department of Physics machine shop.

Shown in the photo is the X-ray source entering from the top of the image. The stage enters from the right on a 4.5" conflat flange. A horizontal shaft turns a mitre gear that engages another mitre gear. The gear ratio is chosen to yield roughly 1 rotation/minute. The white material on the stage is a Macor insulator, allowing one to float the stage or ground the stage through an ammeter. The ion gun enters from the top right at about 2 o'clock.

"Zalar" rotation or rotational depth profiling, RDP, is named after A. Zalar who originally used azimuthal rotation of a sputtered sample to improve the depth resolution of depth profiles using Auger electron spectroscopy. By rotating the sample the cratering due to sputtering is more symmetric in depth profile and thus regions of equal depth are probed by XPS and AES.

Reference: A. Zalar, Thin Film Solids 124, 223. (1985).

Important Change: N2 Venting

An important change: we now have an essentially infinite supply of dry N2 for venting the XPS! Thanks to Bob Smith, the boil-off of the LN2 tank is being plumbed into the building for use in a variety of applications. On the wall to the right of the XPS is the hardware shown in the photo. The yellow valve opens the dry N2 to a regulator shown at the left of the image. The regulator should only be set at a few PSI.

Important: Since there is now a real possibility of over pressuring the XPS intro chamber-- one result being blowing a view port-- unlatch the XPS chamber door before venting! This will prevent over pressuring!

Since this is an effectively infinite source of dry N2 on the timescale one would do anything with the XPS, keep N2 flowing through the intro and out the door while mounting and changing samples. Also, multiple cycles of vent/purge can help get the intro base pressure lower faster.

Thanks to Bob Smith.

Inside the X-ray Source

This is an photo of the XPS X-ray source. The calipers at the bottom of the image indicate 1.0". The square with rounded corners is the anode-- the top is Mg, the bottom Al. There is a little "wall" between the anodes so that electrons from one filament don't accidentally excite X-rays from the opposite anode. The horizontal wires are the filaments. The +15 keV on the anode draws the thermionically emitted electrons from these filaments to the anode. The darker areas on the anodes reflect the regions of the most heat dissipation. Thus, while the X-ray windows reflect the maximum source aperture, these discolored areas more accurately reflect the size of the X-ray source itself.

X-ray Gun Window Replacement

The XPS X-ray source is covered with a thin metalized window. The purpose of this window is to absorb electrons. Electrons are being generated under the window to excite X-rays from the anode, and if these electrons were to hit the sample they could generate core-level and Auger electrons as well as contribute to background. Similarly, electrons ejected from the sample can hit the anode and generate X-rays. This, in fact, is the largest source of spectral contamination from anode "cross talk". In time these windows need to be replaced not only because of heat and radiation damage, but because material is sputtered onto the window during the sputtering process. When that happens the low-energy Al Kα and Mg Kα X-rays are attenuated and there is the possibility for spectral contamination.

Before and after images.

Note: the bar down the middle of the window separates the Al and Mg portions of the anode.

Tuesday, August 23, 2011

Intro Chamber Base Pressure

This image shows the intro-chamber base pressure while pumping with both the turbo pump and the ion pump. With an evacuation pressure in the mid to low 10-7 torr range, the XPS analysis chamber base pressure doesn't move out of the 10-8 torr range during sample introduction, allowing for a quick return to the 10-10 torr range for analysis. The bottom unit is the ion pump controller showing a current of 82 µA.

Please look here for the configuration of the intro chamber. The gauge being read is IG1 and the ion pump is IP.

UHV Intro Chamber + Oven Chamber

The CMMP XPS system has been fitted with a special intro chamber. This chamber replaces the standard PHI 5000 Series XPS intro chamber to provide additional functionality and lower base pressures. The differentially pumped sliding seal intro arm has been replaced with a magnetically coupled intro arm to allow for lower intro base pressures. The intro chamber is also fitted with both a convectron as well as an ion gauge to complement the TC gauge that is integrated into the 5000 Series XPS automatic valve control. A residual gas analyzer (RGA) allows for intro chamber vacuum diagnostics, and a small ion pump allows the intro chamber to stay evacuated without turbo pumping and helps bring the ultimate intro chamber base pressure to UHV levels-- 4 x 10-7 torr. Gas service is provided so that samples can be exposed to gasses for catalysis and surface chemistry studies. The intro chamber connects to an ancillary oven chamber. Samples can be brought from the XPS analysis chamber into this oven chamber for heat cleaning. The oven chamber can be isolated from the analysis chamber and differentially pumped through the intro chamber.

The images that follow show the intro and oven chambers from different vantage points. The labels apply to all images.

IC is the intro chamber while OC is the oven chamber. They are connected by a cross-over path XO and separated by a manual gate valve SV. The intro chamber is pumped by the turbo pump through the pumping line IPL. The pneumatic valve V1* separates the intro chamber from the XPS analysis chamber. The "*" designates a pneumatic valve. These pneumatic valves are controlled through the 5000 Series XPS automatic valve controller. All of the automatic valve controller designations still apply: V3* opens to the turbo pump; V2* to the N2 vent gas; and V4* the differential pumping of the ion gun. V3*, V2* and V4* are all beneath the table-top and are not shown in these pictures.

Samples are introduced through a door, D, which must be opened manually. AI is the intro chamber sample introduction arm, and AO the oven chamber sample introduction arm. AI allows for the rotation of the sample fork. The tilt and lateral positioning of AO is adjusted through an XY-adjustment labeled XY. This must be carefully adjusted to prevent sample stubs from being caught in the oven. The oven is isolated from the XPS analysis chamber using valve OV. OFT is the feed through for current and thermometry to the oven.

Gauging is performed with a Granville Phillips model 307 controller (not shown). IG1 is the ionization gauge read as gauge "1" on the controller, and CVA is the convection gauge read as gauge "A" on the controller. This combination of gauging allows a huge dynamic range of pressure to be covered ranging from atmosphere (780 torr) to ~ 10-10 torr.

RV is the roughing valve. This allows one to open the intro chamber, IC, directly to the roughing pump. This allows for a thorough rough evacuation before the auto valve controller opens V3* (not shown) to the turbo pump. RV-S is a mechanical switch that closes the up to air valve V2* (also not shown) when RV is opened. This is a work around that allows us to integrate this new intro chamber with the existing auto valve controller. RV opens directly to the rough pump and allows the intro chamber to be pumped through the roughing line RPL. In some of these images the ion gun differential pumping line, IGPL, can be seen.

IP is a small ion pump. This ion pump is not gated from the intro chamber, but quickly disgorges physisorbed gasses in the Penning cells when it is turned on. It's function is to keep the intro chamber in the 10-6 torr range when V3* is closed and the turbo is not pumping on the intro chamber-- and to help pull the intro chamber base pressure to the low 10-7 torr range when both the turbo and ion pump are working together.

The residual gas analyzer, RGA, can be gated from the intro chamber using the valve RGAV. The RGA performs mass spectroscopy on the intro chamber allowing one to diagnose problems with gas evolution from samples. It can also be used to examine the evolved gasses when samples are heated in the oven. The main gate valve V1* can be opened and one can also perform RGA on the XPS analysis chamber.

Any operation using the pneumatic valves through the automatic valve controller is protected so that the user can not make a serious error. That is not the case with the manual components of the intro and oven chambers. Serious damage is possible with misuse.

Please refer to the images in this post when procedures using the intro chamber and oven chamber are presented in the future.

To the credit of the CMMP technical staff, specifically Ian Winger and Jim Valentine, this enhancement of the 5000 series XPS was entirely designed and built in house from existing UHV hardware.

Wednesday, August 17, 2011

XPS Base Pressure

This is a picture of the XPS ion gauge controller showing a good but typical base pressure: 1.6 x 10-10 torr. From simple kinetic theory, if every molecule that hits a surface sticks, then the time to form a monolayer is: t = 2.25 x 10-6/P there t is in seconds and P in torr. At this base pressure the monolayer formation time is ~ 3.9 hours-- long enough to prepare a clean surface and study it with XPS. In practice, at room temperature the sticking probability is much less than one so typical monolayer formation times are much larger.

Argon from Sputtering

Sputtering in XPS is performed with Ar+ ions. In some rare cases they can be embedded in the matrix being sputtered. Here we see Ar 2p XPS photoelectrons in an Ar+ sputtered In foil. This peak has a P/B of ~ 0.002.

Tuesday, August 16, 2011

InOx: Wagner Plots

The high energy resolution of XPS, < 1 eV, allows one to identify chemical states through chemical shifts-- shifts of the core level XPS binding energies due to the charge transfer that occurs in the binding of different chemical states. This has been demonstrated in the case of AuOx surface oxidation elsewhere in this blog.

In some cases the shift of XPS lines is insufficient to identify chemical states. In the case of indium metal, the In 3d5/2 XPS BE is 443.8 ± 0.1 eV based on the entries in the NIST XPS database. The corresponding In 3d5/2 BE's for In2O3 are 444.8 ± 0.2 eV. In(OH)3 and In(OH)3*nH2O have a In 3d5/2 BE of 445.1 ± 0.1 eV. Metallic indium can be differentiated from the indium oxide and hydroxide phases, but the oxide and hydroxide chemical states can't be differentiated using the XPS BE chemical shift alone.

The Wagner plot for In, In2O3, In(OH)3 and In(OH)3*nH2O is shown. The Wagner plot shows the binding energy of the XPS In 3d5/2 peak on the X-axis and the kinetic energy of the Auger In M4N45N45 peak on the Y-axis. The diagonal lines are lines of constant modified Auger parameter-- the sum of the XPS binding energy and Auger kinetic energy. Each chemical state is indicated with horizontal and vertical lines indicating the range of the available data-- not the uncertainty in the available data.

What is immediately evident is that chemical states that might be difficult to identify through small chemical shifts in the XPS 3d5/2 binding energy-- might be easy to identify through huge changes in the Auger M4N45N45 kinetic energy, as in the case of In and In2O3. Also phases with little difference in the XPS 3d5/2 binding energy, such as In2O3 and In(OH)3, have a much more significant chemical shift in the Auger M4N45N45 kinetic energy. In(OH)3 and In(OH)3*nH2O can only be differentiated by a shift in the M4N45N45 Auger line.

A heavily oxidized In foil was sputtered while performing XPS. The second image shows the In 3d doublet. Sputtering immediately increases the strength of the In 3d5/2-3d3/2 doublet as the C 1s signature of surface organic contamination is greatly reduced. It should be noted that through very aggressive sputtering-- up to 5 µA at 3 kV-- the In 3d doublet increases in strength, but the BE of the 3 d5/2 peak shifts no more than -0.5 eV lower in binding energy. Because of uncertainties in tabulated XPS values this alone says very little about the surface indium chemistry. The Wagner plot above shows that little difference in XPS BE of the 3d5/2 should be expected between metallic indium and oxide/hydroxide chemical states-- no more than 1 eV and possibly less depending upon which tabulated values are considered "standards".

The last image shows the In MNN Auger region. The Auger structures, like the In 3d XPS doublet, is clearly visible only after surface contamination is removed by sputtering. The M4N45N45 peak is the one considered in the Wagner plot above. The kinetic energy of the M4N45N45 Auger peak stays unchanged through most of the sputtering at ~ 406.5 eV, though the strength of the Auger structures does increase consistently as the surface is cleaned. The very weak XPS In 3s peak also becomes more prominent. After very aggressive sputtering the M4N45N45 Auger peak kinetic energy increases to about 410.0 eV.

The Wagner plot shows the change of ~ -0.5 eV in In 3d5/2 BE and + 3.5 eV in In M4N45N45 KE to be consistent with an In2O3 surface being cleaned to a metallic In surface. The initial In 3d5/2 BE was around 445.0 eV with an In M4N45N45 KE of around 406.5 eV. The Wagner plot identifies this clearly as one of the indium oxide/hydroxide chemical states. The larger Auger KE tends towards the In2O3 oxide state since the hydroxides have smaller Auger kinetic energies with a similar XPS binding energy. The final In 3d5/2 BE around 445.0 eV and M4N45N45 Auger KE around 410.0 eV is entirely consistent with metallic In.

The purpose of this demo is to demonstrate that XPS peak chemical shifts alone are often not enough to elucidate surface chemical states. Auger peaks are often more sensitive to surface chemistry than XPS peaks, and the combination of the two as demonstrated in the Wagner plot is often the key to examining surface chemistry.