This is a wonderful explanation of FEG operation from SEM Geologist!

It also would be great to hear from JEOL EPMA FEG users what they have found on JEOL instruments...

Has anyone had a go at making any synthetic Fayalite for distribution or found a source, since above

We have both SX100 and SXFiveFE and so I can see fundamental differences between using W-hairpin and Shottky's emitter. They require absolutely different analytical strategies, on one hand "Field Emission" opens new horizons, on other hand it brings pretty nasty not advertised disadvantages. Because of these instabilities we were running our SXFiveFE at constant 15kV, and EHT never OFF. Rarely switching to 25kV for sulphide analytical week. (stability within 0.3% of set beam current value). As for beam current drifts we had experienced such, but not so severe. (rather stabilized in 30min instead of 4h). That requires different analytical approach polling similar condition requiring analytical works into longer sessions.

I actually am not surprised about long stabilization at all (I voted "I am using Cameca FE-EPMA", and as I couldn't check two positions in the poll, for stabilization time it would be "<1h" - but that is such a slippery slope where it is easy to go also into >12h (or one more position could be added named as "never"). What I am surprised actually that this is brought to wider attention so much late. This aspect of Thermally Enhanced Field emission (known as Shottky's emitters, further abbreviated by me as TFE) of electron beam is completely left out in all SEM handbooks. For thin-film analysis simple W-hairpin emitter is the only sane option.

As for Cameca FE machines, so far from my 12 year experience with such instrument - for 3 months after new emitter installed it would be able to stabilize in less than 20min after switching conditions. However, if being run by Cameca handbook and protocols such tip would start deteriorate and could reach the state where stabilization falls into "never". Pushing Cameca service aside, and running with improved protocols preventing tip deterioration, could keep switching condition stability <30min, even with 2 year old tip. Then there are other variables which influence beam drift after condition switching - contaminated flash aperture, which in case of Cameca instruments are not easily accessible, and thus In my case was not replaced for 5 years for life-time of TFE.

Why it takes so long to stabilize the beam current after changing the HV?
Because TFE is not thermionic emitter, but field emitter directly depending from electric field which changes with any change of EHT (we have two overlapping fields: between emitter and Extractor, and between emitter and Annode - that is surprisingly somehow completely missed from all practical handbooks and general TFE knowledge) - The Extractor's value stays the same when switching between the EHT conditions (that is applied voltage between emitter and extractor), but the field affecting the tip actually changes as it is a sum of mentioned fields!

That was the easiest part to grasp. Then lets go to more complicated part: 4 horses of apocalypse, or 4 variables which all depends from each other – emission, temperature, field, geometry.

Geometry of tip
Jeol and Cameca EPMA uses TFE designed for SEM work - they are designed to achieve sub 10nm beams, and thus the diameter of tip surface often is about 300-200nm (the small beam diameter is achieved with utilization of virtual source, which is kind like small point within the tip of emitter, where described few hundred nm tip surface acts like kind of projecting "lens"). This geometry reacts to field changes and is able to erode atom by atom from emitting facet and recrystallize those at its sides increasing diameter, or changing its shape (from round into square/Rhomboid). Emission is not uniform from this tip surface - normally it is weakest at middle, and highest at rims, and for SEM/EPMA work we normally use the middle emission part discarding outer parts with help of numerous apertures and condenser lenses... unless column is badly aligned or optical path is physically bent making it impossible to align the emission across all conditions. After changing the field (i.e. changing EHT) the previous geometry becomes unstable and it starts to change for least energy in the newly set field. The pace of that crystallization depends proportionally to the temperature of the tip and geometry itself. I.e. if tip end diameter grows above 1µm (or starts like that) it becomes less affected by field and temperature (That is why Shimadzu EPMA-FE use such specifically tailored TFE for EPMA, not SEM).

Temperature of the tip.
Default old-knowledge based SEM TFE datasheets defines the optimal heat current to give about 1800K (that is for new tip) - the temperature which allows to produce less noisy emission, while still preventing the TFE from complete disintegration - temperature optimal for SEM/TEM high resolution, low noise applications. Setting temperature above would disintegrate tip, where setting it below would start introducing white noise. The problem starts with aging TFE, as it loose mass (mostly through loss of H2O and OH- from ZrOx reservoir, and Zr itself) and the heat current I_new from datasheet (for 1800K) would make higher tip temperature exceeding that (optimal) 1800K (less mass to heat up with same current). Many SEM OEMs ignore that (or are unaware) and that basically kills TFE tips in 1-3 years. Correct heat current decreasing protocol depending from tip aging is hard to implement as there are no way to measure TFE temperature directly (which is different from production setup, where datasheets for I_heat vs T_emitter are defined from direct temperature measurements). And then there is environmental factor - Datasheet I_heat/T_emitter != I_heat/T_emitter on SEM/EPMA. On SEM/EPMA when setting initial 1800K I_heat current values from datasheets tends to actually set about 10 degrees more, as SEM and EPMA have other heat sources heating the casing of EPMA/SEM itself (especially Cameca SX line with its 50 °C stage always-power-on motors). Also Air conditioning can swing these temperatures (AC failure could lead to catastrophic failure of TFE when run at 1800K. Also there is huge thermal capacity of metal case of SEM/EPMA, which adds few more degrees up when TFE is working for at least a week.

Emission of the tip
Emission is proportional to geometry, temperature and the field. It makes it absolutely counterintuitive when we observe changes in emission made by change of the field. In most of case if we see direct response of increase of emission when changing the field, we can be confident that it will drift to lower in long term (few hours, days). And otherwise - direct decrease of the emission after changing the field could be observed to drift to even higher emission than initial after some hours or days. There are few loop-feedbacks of these cause and effects - emission also influence of tip temperature by Nothingham's effect - higher emission means more cooling, lower emission means less cooling by that effect. Take this as lax example: less cooling -> higher temperature -> larger elongation of tip -> emission surface brought closer to extractor -> field increase -> emission increase -> more cooling -> emission decrease -> less cooling...
So on one hand Emission depends from geometry, temperature and field, but also it influence these variables.


So what could be done to decrease these instabilities?:
A. decrease the I_heat to bring it 30 - 40K below 1800K - in case of new tip it is easy as TFE datasheet can be used for that. It rather wont stop geometry changes but would slow it down significantly making it very slow drift up/down or down/up where any beam current regulators should cope with such slow drifts. For aged tip it can be rather tricky as most likely tip is already running at 1840K and further compensation of I_current would be needed.
B. decrease the I_heat to bring it 100K lower (i.e. 1700K) - that will produce more noisy images, albeit introduced random white noise wont affect the analytical work, but should stop recrystalization of tip completely.
C. Ditch SEM tailored TFE and install EPMA tailored TFE with at least 1000nm tip diameter, set it to around 1790K and forget.
D. Do nothing, let the Tip deteriorate (shorten up the tip and widen up diameter) - the observed instabilities between EHT conditions will disappear, emission will drop significantly, the noise and other source of drifts will dominate the stability issues, so that EHT cross condition instability will become insignificant issue completely burried to other instabilities (20% and more of set value).

(experimental)
E. Increase extractor to its max (6kV? 7.2kV?) so that tip would get one and only one shape of highest  possible field (extractor field would dominate, and EHT field would be insignificant) (that albeit increases the risk of catastrophic failure of Arching in case of Vacuum problems). This would allow to use SEM TFE which would keep the same shape across changing EHT and also should eliminate the setting time completely. This however tend to add new source of noise at least on Cameca SXFiveFE, which I still can't simply find the explanation. It could be HV power supply depending issue.

Chose your poison wisely  .

Can SXES be quantified in DTSA-II? Loading standard spectra and unknown spectra? I guess spectra files might need to be converted into readable format unless it produces EMSA? But theoritcally would it work?

I wanted to bring this question up on whether the standard used for the blank correction needs to have a matrix match with the unknown or not.

Because it depends on what exactly is causing the background artifact. For example, if the background artifact is in the Bragg crystal (secondary Bragg diffraction artifacts, e.g., PET crystal for Ti Ka) or the gas detector (absorption edge artifacts, e.g., Ar absorption edge), the zero blank matrix may not matter all that much.

But if the background artifact is related to the sample composition, e.g., measuring off-peak bgds across an absorption edge caused by an element in the sample, the blank matrix could matter a lot.

Clearly if we have a perfect matrix match blank standard that would be ideal in all circumstances. But exactly how much is needed depends on the cause of the background artifact.

This all reminds me of a discussion Mike Jercinovic had regarding creating a synthetic monazite blank:

https://smf.probesoftware.com/index.php?topic=928.msg8506#msg8506

Working with Peng Jiang in Hawaii we ran some major elements on EDS (Bruker Esprit) and minor elements on WDS. Here is an analysis of a natural garnet as a secondary standard:

St  102 Set   1 Garnet, Verma (Mn), Results in Elemental Weight Percents
 
ELEM:       Si      Fe      Mg      Ca      Mn      Al       O
TYPE:     ANAL    ANAL    ANAL    ANAL    ANAL    SPEC    CALC
BGDS:      EDS     EDS     EDS     EXP     LIN
TIME:    39.98   39.98   39.98   60.00   60.00     ---     ---
BEAM:    19.84   19.84   19.84   19.84   19.84     ---     ---

ELEM:       Si      Fe      Mg      Ca      Mn      Al       O   SUM  
   112  16.589  15.708    .000    .193  19.249  11.020  38.886 101.645
   113  16.648  15.719    .001    .183  19.305  11.020  38.970 101.846
   114  16.786  15.698    .035    .171  19.238  11.020  39.119 102.067
   115  16.757  15.654    .050    .181  18.991  11.020  39.016 101.670
   116  16.821  15.710    .038    .185  19.045  11.020  39.113 101.931

AVER:   16.720  15.698    .025    .182  19.166  11.020  39.021 101.832
SDEV:     .098    .026    .023    .008    .138    .000    .099    .178
SERR:     .044    .011    .010    .004    .062    .000    .044
%RSD:      .58     .16   91.83    4.30     .72     .00     .25

PUBL:   17.240  14.020    n.a.    .170  19.050  11.020  39.077 100.577
%VAR:    -3.01   11.97     ---    7.31     .61     ---     ---
DIFF:    -.520   1.678     ---    .012    .116     ---     ---
STDS:      114     103     103     114     303     ---     ---

I cannot speak to the accuracy of this particular (natural) standard, but the analysis looks OK, but the Fe on EDS is way off.

Then we added Mn as an interfering element because Probe for EPMA can perform interference corrections whether the element is a WDS or an EDS element.  We utilized a synthetic Mn2SiO4 standard for the interference correction because it contains a known amount of the interferring element (Mn) but none of the interfered element (Fe):

St  102 Set   1 Garnet, Verma (Mn), Results in Elemental Weight Percents
 
ELEM:       Si      Fe      Mg      Ca      Mn      Al       O
TYPE:     ANAL    ANAL    ANAL    ANAL    ANAL    SPEC    CALC
BGDS:      EDS     EDS     EDS     EXP     LIN
TIME:    39.98   39.98   39.98   60.00   60.00     ---     ---
BEAM:    19.84   19.84   19.84   19.84   19.84     ---     ---

ELEM:       Si      Fe      Mg      Ca      Mn      Al       O   SUM  
   112  16.573  14.191    .000    .193  19.279  11.020  38.443  99.699
   113  16.632  14.199    .001    .183  19.335  11.020  38.525  99.895
   114  16.770  14.182    .035    .171  19.268  11.020  38.675 100.122
   115  16.741  14.158    .050    .181  19.021  11.020  38.578  99.749
   116  16.805  14.210    .037    .185  19.074  11.020  38.674 100.005

AVER:   16.704  14.188    .025    .183  19.195  11.020  38.579  99.894
SDEV:     .098    .020    .023    .008    .139    .000    .100    .175
SERR:     .044    .009    .010    .004    .062    .000    .045
%RSD:      .58     .14   91.83    4.30     .72     .00     .26

PUBL:   17.240  14.020    n.a.    .170  19.050  11.020  39.077 100.577
%VAR:    -3.11    1.20     ---    7.48     .76     ---     ---
DIFF:    -.536    .168     ---    .013    .145     ---     ---
STDS:      114     103     103     114     303     ---     ---

Much better!

I am surprised that Peng's FEG takes so long to stabilize.

I know that Cameca had a lot of trouble with their early FEG but I'd never heard of any stability issues with JEOL's.

In addition to switching beam currents for samples with both major and trace elements, there's the multi-voltage method for thin film analysis:

https://smf.probesoftware.com/index.php?topic=111.0

Having a stable beam after switching beam current and/or beam voltage is pretty darn essential.

With our Cameca W gun at UofO, there are zero stability issues with changes in beam current or voltage. Though early on we had relatively short filament life, but once the column has run for six months or so, the filament life improved significantly.

If you do have a FEG instrument, please vote in Peng's poll at the beginning of this topic.

Aloha EPMA folks,

Yesterday I discussed the beam stability of our newly installed JXA-iHP200F at the University of Hawaiʻi at Mānoa with John Donovan. He was surprised by the long waiting time (8–12 hours) for the beam to stabilize and suggested that I make a post here.

I also think that a ~12-hour stabilization period is quite long, even though this is a new emitter and a new instrument. I'm curious how other newly installed JXA-iHP200F probes or other JEOL probes have behaved in terms of beam stability and overall stabilization time. I greatly appreciate your experience and thoughts

Below is a quick summary of how our new probe beam behaves:
1. When switching from a high accelerating voltage to a lower one (e.g., 20 kV to 10 or 5 kV), the beam current increases rapidly during the first 8–12 hours, followed by a much slower increase thereafter as it gradually stabilizes.
2. When switching from a lower accelerating voltage to a higher one (e.g., 5 to 10 kV, or 10-15 to 20 kV), the beam current first increases rapidly and is then followed by a gradual decrease, stabilizing around 12 hours.

After stabilized, the beam variation was < 1% within 12 hours.

You cannot view this attachment.
- Above was a test with the voltage switched from high (15-20kV) to low (5kV).

You cannot view this attachment.
- Above was a test with the voltage switched from low (5kV) to high (10kV).

JEOL engineers suspected that room temperature might be affecting low voltage beam stability. So we added a temperature logger and showed that temperature was not the factor (image below).

You cannot view this attachment.
- Above was a test under 10kV, 50nA, for 20h. No significant temperature change but beam dropped and it took about 12 hours to stabilize.

I also noticed that, even under the same voltage (stabilized for a few days), changing beam current also appeared to "trigger" the stabilizing period.
You cannot view this attachment.
- Above was a test under 10kV 50nA for 36h. Beam current was changed from 100nA (used for X-ray mapping) to 50nA (for beam stability test). It took about 4 hours for beam to stabilize.

Heather Lowers of the USGS Denver had a nice suggestion when using the statistically significant format output feature:



That is, when a concentration is below the detection limit, instead of just printing "n.d." for not detected, print a "<" symbol before the detection limit.  Like this:

St  263 Set   3 Fe2SiO4 (synthetic fayalite), Results in Elemental Weight Percents
 
ELEM:       Si      Mg      Mn      Fe       O
TYPE:     ANAL    ANAL    ANAL    ANAL    SPEC
BGDS:      LIN     LIN     LIN     LIN
TIME:    60.00   60.00   60.00   60.00     ---
BEAM:    30.01   30.01   30.01   30.01     ---

ELEM:       Si      Mg      Mn      Fe       O   SUM  
  1611    13.6   <.011   <.020    54.5  31.407   99.51
  1612    13.7   <.010   <.021    54.6  31.407   99.66
  1613    13.7   <.011   <.020    54.6  31.407   99.68
  1614    13.7   <.010   <.020    54.5  31.407   99.64
  1615    13.7   <.011   <.021    54.7  31.407   99.72
  1616    13.7   <.010   <.021    54.5  31.407    99.6

AVER:     13.7   <.010   <.020    54.6  31.407  99.635
SDEV:     .025    .005    .012    .059    .000    .074
SERR:     .010    .002    .005    .024    .000
%RSD:     .180-214.108-558.567    .108    .000

PUBL:   13.785    n.a.    n.a.  54.809  31.407 100.001
%VAR:     -.86     ---     ---    -.44     ---
DIFF:    -.119     ---     ---   -.243     ---
STDS:      256      12      25     395     ---

Cool idea!

Update Probe for EPMA from the Help menu and you will have this new feature.

Quote from: Ben Buse on August 11, 2026, 08:56:34 AMDoes anyone have any synethic foresrite to share Mg2SiO4, I see you mention it both in above post and above M&M poster.

I also see Peter McSwiggen presented on peak shifts in Mg system - using 2nd order line - so more visible
https://academic.oup.com/mam/article/32/Supplement_1/ozag053.836/8742788

Ben is referring to this post and attached poster presentation pdf:

https://smf.probesoftware.com/index.php?topic=1831.msg14457#msg14457

This synthetic Mg2SiO4 material is available from this company:

https://www.opt-oxide.com/en/