View Source

Step-by-step guide - Femtosecond (fs) timing in sample chamber 1 (SC1) at CXI.

Setup

Check all diodes at the beginning of the shift (ideally this should also be done a few days before the experiment to give you time to fix things that don't work), by plugging them into Lecroy and visual check (with bias-T).

Install a fast diode and mirror (roughly at 45 degrees with respect to the beam) on the SC1 post_sample_x stage (CXI:SC1:MMS:02.RBV) downstream on the interaction region. There is a danger of driving this stage into the Jungfrau or other components in SC1, so it might be disabled. It is also good practice to disable the motor when not in use.

The fast diode has a small area and therefore can be difficult to align. Ideally, a Y stage (PP-30 piezo) is used to allow remote motion of the diode to align to the laser and x-rays.

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > Mirror-and-Diode-samplex-stage-outside.jpeg

Due to the limited travel range of the long X-stage in SC1, the diode and mirror must be installed as far as reasonable towards the -X side, overhanging away from the carriage

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > Mirror-and-Diode-samplex-stange.jpeg

Roughly align the fast diode and mirror using the reference laser at the time tool.

(-4 mm in position but this will soon be changed to 0 mm. If the "out" position is 92 mm, then -4 mm, if 96 mm, then 0 mm).

Install a regular diode connected to the Acqiris on the sample chamber door and align the reflected reference laser beam on the mirror onto that diode.

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > External-Diode-SC1.jpeg

Repeat with the fs laser and maximize the Acqiris signal by moving the diode on the chamber door. An ND filter may be necessary to reduce the signal, as shown in the side-on view below

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > External-Diode-SC1-sideview.jpeg

Also install a fs laser diagnostic diode (OPC-TP) and connect it to the Acqiris. This is useful to confirm that the fs laser is firing during the experiment.

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > image2022-2-1_20-20-10.png

Insert a nozzle rod with both a clear YAG and a frosted YAG.

At CXI, the clear YAGs are typically thinner, around 20 um, whereas the frosted YAGs are thicker, around 50 um

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > YAG-stick-holder.jpeg

If doing the timing with the shroud on, the bracket needs to be facing in the -Z direction when inserted into the sample chamber or it will collide with the exit cone.

Determine the position of the X-ray spot at the IP using the clear YAG and the beam with 1x10-4 Transmission (although the beam can be seen with less intensity). Mark the position of the beam with crosshairs (cyan and magenta in the example image below).

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > focused-beam-at-SC1-IP.jpeg

Zoomed-in view:

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > focused-beam-at-SC1-IP_zoomed-in.jpeg

Rough timing

If timing is desired to be done at atmospheric pressure in SC1, then the MPS interlock must be bypassed to allow the beam to come to CXI with the detector gate valve close. this requires bypassing Link Node 40, Card 3 Channel 15 to open for the duration of the timing measurement at atmosphere. This bypass can only be enabled by ACR and you must call them to request this at x2151.

Move the fast diode into position (at this point should only need to move the SC1 sample_x stage) and move the IP YAGs out of the beam.

Align the diode to the x-ray beam first since the x-ray beam cannot be moved. The laser will later be steered to the diode left int he fixed location of maximum x-ray signal
The focused x-ray beam may damage the diode. If the signal on the diode is too weak for 10^-4 transmission or less, it may be that the diode is improperly biased.
Increased x-ray signal may be achieved without damaging the diode by making the x-ray beam bigger using the DG2 Be lenses as described later in the fine timing section.
Make sure the Lecroy is triggered off the EVR.

Measure the laser signal on the diode using the LeCroy scope to make sure there is overlap

Shutter the fs laser and unshutter the X-rays. Move the diode around in x and y to maximize the X-ray signal (if the steps are very large then the laser may need to be re-pointed)

Click Store on the Lecroy to save a trace of the X-rays on the scope

Shutter X-rays off and unshutter fs laser.

Temporally overlap the X-rays and the fs laser.

If needed, align the fs laser to the diode (usually not needed). Move laser delay time (LAS:R52B:EVR:31:TRIG0:TDES) to bring laser within 100 picoseconds of the X-rays using the scope. Alternatively, change the Target Time in the fs laser window. Align the fs laser trace so that the leading edges of the traces line up. You can center the diode signal on the Lecroy by changing the Delay. Try to match the amplitude of the X-ray and fs laser signals. Type the time below the ns Offset in the fs laser window in the ns Offset box, which will reset the Target Time to 0.

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > laser-timing-evr-window.jpeg

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > lecroy.png

Fine timing

Once the laser is roughly timed, it is time to move onto the finer timing measurement using a cross-correlation measurement.

Background Information

See the Time Tool Confluence page for more details about how the time tool works.

Fine timing is done by using the photoelectrons generated by the interaction of the X-rays with a material to effectively turn a non-metal into a metal (sea of electrons produced from photoelectrons) that changes the index of refraction and, more importantly, the reflectivity and transmission of the material. If an optically transparent material, such as Si3N4 or a clear YAG, is irradiated with an optical laser, the vast majority of the photons will be transmitted through the material. However, if the X-rays arrive before the optical laser, the non-metal to metal transition (effectively) caused by the photoelectrons will increase the reflectivity of the material and decrease the transmission of the optical laser through the material. This effect can be measured either through a decrease in the transmission of the optical laser or through the increased reflection of the optical laser. In the case of fs timing at CXI we use a measurement of the decrease in transmission of the optical laser (SC1 measurement) or white light produced by the laser (time tool, done using a portion of the optical laser diverted from the sample chamber).In SC1, the change in transmission is measured with a simple diode but the diode cannot be put directly in the beam path after the target since both the x-rays and optical laser beams would hit that diode. Therefore, the 45 degree mirror which sends the beam to another diode on the chamber door is used to separate the optical laser from the x-ray beam. The mirror reflects the laser and the x-rays are absorbed. It is OK to dump the full x-ray beam on this mirror. Even if it damages slowly with time, the mirror is cheap.

Align the laser with crosshairs on the inline cameras that represent the location of the X-ray beam.

In the image below, the bright spot is the laser spot at the interaction point (IP) visualized using a frosted YAG. The frosted YAG is on the upper half of the image. There is a small gap followed by the bottom 1/4 of the image having a clear YAG in this particular instance but it is likely that the mounting of the 2 YAGs will vary with changing experiments and if Ray Sierra broke them or not...

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > misaligned-laser_SC1.png

The laser pointing is changed using Pico X (CXI:LAS:PIC:03) and Pico Y (CXI:LAS:PIC:02). These PVs can be changed and which of the motors does pitch and yaw can change so these are not set in stone.Trial and error is required to verify the direction of motions.

Enlarge the X-ray spot size using DG2 lenses.

When the laser is roughly centered on the X-ray spot (the laser should be on the order of 100-150 um but this can vary between experiments with larger spots being possible), the X-ray beam needs to be made similar in size at the interaction point to the laser to maximize the signal, to make sure that the part of the target the laser sees is fully modified by the x-ray beam. In order to do this at CXI, insert Be lenses at the DG2 location to focus the beam upstream of SC1 and produce a larger, divergent beam at the IP.

The DG2 lens screen is found in the DG2 section of CXI home. It is password protected with the usual CXI password.
CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > DG2-lenses_screen.jpeg

The Be lenses will need to be aligned in both x and y using the DG2 YAG to monitor alignment. Then, the DG2 slits need to be narrowed such that there is no beam intensity outside the lens aperture. If intensity remains outside of the lens aperture it will be focused to the 1.3 um spot size in the IP and damage the YAG at full intensity.

Beam on DG2 without lenses:
CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > Beam-on-DG2-no-lenses.jpeg

Beam on DG2 with lenses inserted but without slits narrowed:
CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > Beam-on-DG2-with-2x50um-lenses.jpeg

Beam on DG2 with lenses inserted and with DG2 slits narrowed:
CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > Beam-on-DG2-with-2x50um-lenses_DG2-slits-closed.jpeg

Open all gate valves and remove objects upstream of SC1 before bringing the beam to the IP. Since the beam is being refocused upstream of SC1 to make the larger beam, we could inadvertently place the focused beam on upstream diamond windows or other objects and damage them

Look at the beam on the clear YAG in SC1 (or the frosted YAG is fine for this as it allows for a check of the spatial overlap of the x-rays and optical laser)

The beam should be similar in size to the laser beam or the signal used for timing in SC1 will be low
CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > prefocused-beam-at-SC1-IP.jpeg

Optional: Use the large beam at SC1 to align the DSB slits, SC1 jaws, and SC1 ap0 and ap1 motors to be roughly centered. It is easy, and the SC1 +Y jaw and DSB +Y slits are particularly useful for cleaning up scatter from the nozzle that may make the data quality worse

Find t0 at the IP with a fine timing scan.

The exact procedure for doing a fine timing scan tends to change often depending on the status of various scripts. Currently (as of Feb. 2022), the icxi script does not seem to work. A manual scan will always work if you cannot get an automated scan to work.

Set up Aquiris

For the timing scan you will need close to 100% transmission in the sample chamber (likely > 25%). This is required because the desired transmission change is driven by high intensity x-ray effects.

The laser signal should be on the Acqiris using the diode installed on the chamber door. In the case of LS87 we were using Channel 1 to look at the diode outside SC1 and we should strive for a peak height of ~1V to ensure that the cross correlation signal is strong.

Before running the scan, a plot needs to be set up that plots the integrated laser intensity on the Acqiris vs the laser ns Target Time (LAS:FS5:VIT:FS_TGT_TIME_DIAL).
CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > SC1-time-scan-PLOTTING-example.jpeg

Set cursors both earlier and later than the laser peak on the Acqiris and in "Expr" under Expression manually add "a integral b" using the integral sign.

Plot A integral b as a function of LAS:FS5:VIT:FS_TGT_TIME_DIAL and choose the hi and low values consistent with the scan. The above example shows the typical final low and hi values chosen when we are very close to t0.

Perform manual scan

Click Reset Plots in the AMI window.

Change the Target Time in the laser timing window by +-1 ps in 100 fs steps. Allow the data to be averaged for a few seconds at each time point. You should get a plot with a clear edge, like this:

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > attachment9.jpg

What is identified as t0 is a never ending matter of debate. However, the majority of the time, the width of the falloff in transmission fairly closely matches the distribution of the jitter in arrival times. Since the measurements made here average many shots, we are sensitive to this jitter. The photoelectron release is assumed to be essentially instantaneous with the width driven by the time difference in propagation of the x-rays and the laser through the material and the jitter in the laser arrival time. Typically, the jitter dominates with a thin target like a 20 um YAG. If the measurement is jitter dominated, then the inflection point is the right choice for time zero. If it is determined that jitter is not the dominant factor in the width of the falloff in transmission, then one can determine a more suitable reference point earlier in time. Since the jitter is typically >200fs and approaching 400 fs, as was the case for the example here, then a 400 fs width of the drop as shown above makes the inflection point likely the right choice. In the example shown above, a t0 value of -200 fs would be a reasonable number with a +/- 100 fs error bar. One thing to be careful is in the past, we have seen "periodic jitter" where the timing would drift back and forth every 20-30 seconds with an amplitude of more than 1 ps sometimes. Under those circumstances, the phase of that "periodic jitter" when the scan is close to time zero could easily move the apparent t0 by a few 100fs. The measurements at every data point of the scan should ideally be longer than any periodic jitter period, if this situation is present on a given day.

Update t0 in the laser timing window:

CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > laser-timing-window.jpeg

In this example, the ns Target Time should be set to -200 fs to determine the value of the final timing number (LAS:FS5:VIT:FS_TGT_TIME, in the above image it is 4392.074100).

The ns Offset should be changed to be the value of LAS:FS5:VIT:FS_TGT_TIME. Once this change is made the ns Target Time should reset automatically to 0.000000.

Once timing has been established in SC1, timing must be found at the time tool. If the time tool delay stage position is roughly known, then timing at the time tool is relatively easy and you can skip to the time tool fine timing section.

Find approximate t0

You have multiple options here. You can set up a “Takahiro” plot. Set up an AMI plot of integrated diode intensity vs. DG2 Wave8 or FEE gas detector intensity. You can also watch the fast timing diode signal on the Acquiris and watch for the signal to decrease when X-rays are turned on.

If using the Takahiro plot, when the X-rays come after the fs laser, these values are uncorrelated. When the X-rays come before the fs laser, there is a negative correlation for these values. Using a binary search, vary the fs laser target time in the Time Tool Plugin window until you find the point where the AMI plot shows a correlated line of points with slope half of the maximum.

Perform am automatic scan with icxi

Load icxi by typing "icxi" in any terminal on the cxi machines (cxi-daq, cxi-monitor, etc.)

Load the time scan by typing "%timescans" into the ipython terminal
CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > setting-up-timescan-icxi.jpeg

For time scan, type "ts.scan_times(start_delay, end_delay, step_size, nevents_per_timestep=360, randomize=True)" in the icxi terminal
CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > time-scan-example.jpeg

The time scan, if done properly, will produce a plot that looks similar to the one below
CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > SC1-fine-timing-LS87.jpeg

What is identified as t0 is a never ending matter of debate. However, the majority of the time, the width of the falloff in transmission fairly closely matches the distribution of the jitter in arrival times. Since the measurements made here average many shots, we are sensitive to this jitter. The photoelectron release is assumed to be essentially instantaneous with the width driven by the time difference in propagation of the x-rays and the laser through the material and the jitter in the laser arrival time. Typically, the jitter dominates with a thin target like a 20 um YAG. If the measurement is jitter dominated, then the inflection point is the right choice for time zero. If it is determined that jitter is not the dominant factor in the width of the falloff in transmission, then one can determine a more suitable reference point earlier in time. Since the jitter is typically >200fs and approaching 400 fs, as was the case for the example here, then a 400 fs width of the drop as shown above makes the inflection point likely the right choice. In the example from LS87 above, a t0 value of -400 fs would be a reasonable number (with -300 fs perhaps being a bit better) with a +/- 100 error bar. One thing to be careful is in the past, we have seen "periodic jitter" where the timing would drift back and forth every 20-30 seconds with an amplitude of more than 1 ps sometimes. Under those circumstances, the phase of that "priodic jitter" when the scan is close to time zero could easily move the apparent t0 by a few 100fs. The measurements at every data point of the scan should ideally be longer than any periodic jitter period, if this situation is present on a given day.

The t0 value would be updated in the laser timing screen
CXI@LCLS > Femtosecond timing at the Interaction Point (SC1 chamber) > laser-timing-window.jpeg

In this example, the ns Target Time should be set to -400 fs to determine the value of the final timing number (LAS:FS5:VIT:FS_TGT_TIME, in the above image it is 4392.074100).