Increasing the laser power incident on the recycling mirrors in the LIGO interferometers

LASER INTERFEROMETER GRAVITATIONAL WAVE OBSERVATORY -LIGO- CALIFORNIA INSTITUTE OF TECHNOLOGY MASSACHUSETTS INSTITUTE OF TECHNOLOGY Technical Note LIGO-T030288-00-W 12/09/03 Increasing the laser power incident on the recycling mirrors in the LIGO interferometers D. Cook and R. Savage This is an internal working note of the LIGO Project. California Institute of Technology LIGO Project MS 51-33 Pasadena CA 91125 Phone (626) 395-2129 Fax (626) 304-9834 E-mail: info@ligo.caltech.edu WWW: http://www.ligo.caltech.edu Massachusetts Institute of Technology LIGO Project MS 20B-145 Cambridge, MA 01239 Phone (617) 253-4824 Fax (617) 253-7014 E-mail: info@ligo.mit.edu C:\User\Rick\Papers\Power_Increase\power_increase.doc 12/22/2003

ABSTRACT Drawing heavily from elog entries made by various authors, we evaluate the present power levels from the 126MOPA laser and through the PSL and IOO optics to the recycling mirror. We then estimate the effort and cost required to increase the laser power and transmission efficiency of the optical components. For reference we use the power levels from the subsystem design documents. However, because these documents are about ten years old and were written before the subsystems were built and installed, we recommend an integrated assessment of the present situation and most cost effective avenues toward increasing the power in the interferometers. Note that we do not address other subsystem requirements such as noise levels, we consider only raw power. 1 INTRODUCTION There are two subsystems responsible for delivering the laser light to the interferometer at the recycling mirror. The light originates in the pre-stabilized laser subsystem (PSL) and is conditioned by the input optics subsystem (IOO) before impinging on the recycling mirror (RM). The relevant components in the PSL are the 126MOPA laser source and the premodecleaner (PMC). In the IOO, they are the electro-optic modulators (EOMs), the modecleaner (MC), and the Faraday isolator (FI). There are a number of other lenses and mirrors in each subsystem, but because these components typically have efficiencies greater than 99% and because none has been identified as a potential source of significant loss of power, they are neglected in this analysis. Simply stated, the design requirement for the PSL was that it deliver 8.5 watts of laser power to the IOO. The IOO was to deliver 6 watts to the RM. We will discuss individual optical components of each subsystem, considering meeting the bottom line requirement of 6 watts at the RM to be paramount. Page 2 of 21

2 POWER BUDGET OVERVIEW An example of a power budget that would meet the requirement of 6 W of laser power at the RM is shown in Figure 1, below. Visibility = 88% Visibility = 95 % T 126 MOPA T PMC PSL T EOMS T MC T IOO FARADAY RM 10.0 Watts 8.5 Watts 6.0 Watts 85% 90 % 85 % 92 % T = 60 % Figure 1 Example of component transmissions that would meet the requirement of 6 watts of power at the recycling mirror The transmissions used in this example are not from optical component design requirements, they simply show one set of parameters that would be acceptable. The 126 MOPA power is the total power in all modes. The specification calls for 11 W in all modes with 10 W in the TEM 00 mode, so this budget would give 10% head room at the front end. 2.1 Present H1 Power Budget An estimate of the present H1 power buget is shown in Figure 2. Explanations of the various power levels and transmissions and references to the measurement data can be found in the relevant sections later in this document. Page 3 of 21

Modematching = 87.5% Visibility = 86 % T 126 MOPA T PMC T EOMS T MC T PSL IOO FARADAY RM 6.15 Watts 4.61 Watts 2.36 Watts 75 % 86 % 64 % 93 % T = 38 % Figure 2 Present (Dec. 2003) esitmate of the H1 power budget with the 126MOPA operating at full power. 2.2 Present H2 Power Budget An estimate of the present H2 power budget is shown in Figure 3. Visibility = 95 % Visibility = 97 % T 126 MOPA T PMC PSL T EOMS T MC T IOO FARADAY RM 6.58 Watts 5.59 Watts 3.04Watts 85.0 % 75 % 88.5 % 82 % T = 46 % Figure 3 Present (Dec. 2003) esitmate of the H2 power budget with the 126MOPA operating at full power. Page 4 of 21

2.3 L1 Power Budget on June 26, 2003 An estimate of the present (Dec. 2003) L1 power budget is shown in Figure 4 1. Note that the 126MOPA was replaced and a new style PMC was installed before these powers were measured. Visibility = 80 %? Visibility =?? % T 126 MOPA T PMC PSL T EOMS T MC T IOO FARADAY RM 7.5 watts 4.56 watts 5.5 watts??? watts 73.3 % 83 %?? %?? % T =?? % Figure 4 Present (Dec. 2003) esitmate of the L1 power budget with the 126MOPA operating at full power. 2.4 Present L1 Power Budget Figure 5 Present (Dec. 2003) esitmate of the L1 power budget. 3 PMC The 126 MOPA light traverses the PMC before reaching the PSL/IOO interface. Thus the laser output power and the PMC transmission efficiency are the two factors that determine the ability to meet the 8.5 W requirement at the PSL/IOO handoff point. Thus the efficiency of the PMC, determines the required power from the 126 MOPA laser. The PMC efficiency depends on the modematching of the laser beam to the PMC cavity normal mode and the ideal transmission of the PMC which depends on the mirror losses and matching of the input mirror transmission to the output mirror transmission plus losses (optimal coupling) 1 Private communications with Rupal Amin. Page 5 of 21

The PMC was recently redesigned to increase the cavity finesse for improved filtering of power fluctations at high frequencies (> 24 MHz) and to facilitate optical contacting of the input and output mirrors to the PMC spacer. One of these new style PMCs has been installed in the LIGO Hanford Observatory (LHO) 4-kilometer (H1) PSL. A second new style PMC has been delivered, but not yet installed in the LHO 2-kilometer (H2) interferometer. 3.1 H1 PMC A new style PMC was installed in the H1 PSL on May 2, 2003 2. The most recent assessment of this PMC s performance was performed on Sept. 18, 2003 3 The performance was evaluated by making a number of power measurements then using the procedure and Matlab program described in LIGO Technical Document LIGO-T010037-00- W 4. The results are: Transmission efficiency = 75% Modematching efficiency = 87.5% (visibility = 87.1%) Estimated average losses per mirror = 325-350 ppm per mirror The mode of the incident laser was not evaluated separately, but the manufacturer specification calls for greater than 90% in a circular TEM 00 mode. Thus 88% modematching efficiency is not far from the 90% best possible if the laser beam quality just met spec. However, the 325-350 ppm losses per mirror are about ten times the expected 30 ppm per mirror. Options for improvement 3.1.1 Improve modematching Experience with the Lightwave 126MOPA lasers indicated that can and often do meet the requirement of 90% of the power in a circular TEM 00 mode. However, they do not typically exceed the requirement, and if they do it is not usually by more than one or two percent. Thus improving the modematching is not likely to improve the PMC transmission by more than a few percent at best. This could be accomplished by adjusting the modematching telescope lens positions and by adding additional cylindrical lenses to circularize the 126 MOPA output beam. Effort and Cost We estimate the effort required to optimize the present modematching telescope and implement the beam circularization optics to be 2-3 days per PSL. The 126 MOPA would 2 http://blue.ligowa.caltech.edu/ilog/pub/ilog.cgi?group=detector&date_to_view=05/02/2003&anchor_to_scroll_to=2003:05:02: 23:38:01-rick. 3 http://blue.ligowa.caltech.edu/ilog/pub/ilog.cgi?group=detector&date_to_view=09/18/2003&anchor_to_scroll_to=2003:09:18: 11:01:02-rick. 4 http://www.ligo.caltech.edu/docs/t/t010037-00.pdf Page 6 of 21

likely need to be moved away from the PMC by 1-3 feet to provide room for the additional optics. The estimate cost is $2k per PMC. 3.1.2 Reduce mirror losses The larger-than-expected losses result in a significant loss of efficiency. The PMC transmission efficiency as a function of loss per PMC mirror is plotted in Figure 6. Figure 6 PMC transmission vs. loss per PMC mirror. The observed mode matching efficiency (M) is 0.875. M=1 corresponds to perfect mode matching. Note that reducing the losses from 325 to 50 ppm per mirror would result in an increase in the transmission efficiency from 0.75 to 0.85. The observed PMC mirror losses may result from operating in air rather than in vacuum. While N. Uehara, the designer of the PSL PMC, had favorable experience operating high-power cavities in air, his tests were likely over durations shorter than weeks or months. All of the measurement at LHO over the past five years have indicated that the PMC mirror losses are hundreds to thousands of ppm per mirror. Several years ago, motivated by the desires both to reduce the PMC sensitivity to atmospheric pressure variations and to reduce mirror contamination, S. Marka, R. Savage, and D. Cook initiated the design of a vacuum chamber for the PMC cavities. This design is approximately 50% complete. Page 7 of 21

Effort and Cost We estimate that the effort required to finish the mechanical design and drawings is about one week, with another one to two man-weeks for procurement, assembly and testing. The cost per PMC is estimate to be about $10k per PMC including ion pump and controller, vacuum windows, etc. This does not include the cost of new PMC cavities. 3.2 H2 PMC The H2 PSL PMC transmission efficiency was last measured on June 13, 2003 5. The measured transmission was 85% and the visibility measured in reflection ([brightdark]/bright) was 95%. The 95% visibility indicated that the beam quality of the 126 MOPA is better than the spec. (90%) so we should expect it to decrease if the laser were to be replaced. While the 85% transmission is better than the 75% measured for the new style PMC on H1, the H2 PMC will have to be upgraded to the new style as well to meet the high-frequency power fluctuations spec. Options for improvement Identical to those for the H1 PMC, described above. 3.3 LHO Spare PMC We do not present have a spare new style PMC at LHO. Options for improvement 3.3.1 Procure and test spare PMC. Effort and Cost Ask Peter King to supply cost. The effort required to test the spare PMC at LHO would be several days. We would have to configure the test stand in the optics lab to allow operation of the PMC servo and we would need the HV power supplies, etc. We may have spares that could be utilized. Perhaps there is a test setup at Caltech that is already configured to enable testing. 3.4 L1 PMC The LI PMC transmission efficiency was measured on June 26, 2003 to be 65%. It appears that this was and old style (lower finesse) model. The PMC may have been replaced by then, but no record of the transmission could be found in the elog. We assume that the present PMC is the new, higher-finesse style. 3.5 LLO Spare PMC We have no info. regarding the existence of a spare PMC at LLO. 4 126MOPA The 126MOPA output power required to meet the PSL requirement of 8.5 W at the PSL/IOO interface depends on the transmission of the PMCs. If we assume that the PMC mirror losses 5 http://blue.ligowa.caltech.edu/ilog/pub/ilog.cgi?group=detector&date_to_view=06/13/2003&anchor_to_scroll_to=2003:06:13: 10:06:08-rick Page 8 of 21

can be improved by operating in vacuum and that the modematching into the PMCs is optimized, it seems reasonable to expect 85% for the PMC transmission. The required 126MOPA power in the TEM 00 mode would then be 10 W, the design requirement. Assuming 90% of the output power is in a circular TEM 00 mode (again, the design requirement M 2 < 1.1), the required total output power is 11 W. Our experience with Lightwave to date indicates that the Gentec power meter they use overestimates the power by about 10% 6. Thus, what Lightwave calls 10 W is actually about 9W. Thus the required total output power when using a PMC with 85% transmission is about 12.4 W ( x0.9 for the power meter x 0.9 for the M 2 -> ~10 W). 4.1 H1 126 MOPA The H1 126 MOPA has been operating continuously since about March, 2001, about 25,000 hours. The NPRO was replaced in May, 2001 7. On Sept. 18 th, 2003 we increased the current of the 126MOPA from 24.4 A to the maximum recommended by the manufacturer, 25.75 A. This resulted in about a 1.25 W increase in the output power 8. Without making any effort to realign the laser optical train, increase the NPRO power, or optimize the NPRO or amplifier pump diode temperatures, we measured 6.15 W of total output power. To meet the 12.4 W level would we would have to realize a factor of two increase in the output power. Options for improvement We have little experience with major realignment of the 126MOPA lasers and what little experience we have is not particularly encouraging. It is unlikely that simple realignment and optimization will result in a factor of two increase in the output power. Thus this laser will likely require an overhaul that includes rebuilding the amplifier (new pump diodes). 4.1.1 Rebuild amplifier LLO has recently returned a 126MOPA laser for an overhaul that includes replacing the pump diodes in the amplifier. The overhaul will include realignment an returning the laser to the original power output specifications (11 W total power with at least 10 W in the TEM 00 mode. Note that these are powers measured at Lightwave, they should be de-rated by 10% because of the calibration of their power meter). Effort and Cost Based on the LLO rebuild estimate, this should cost about $30k. Jim Kayser from LTW estimates that they can turn around a laser in 3 weeks. If the ifo. cannot be without a source for this period, the spare laser would have to be swapped in. At LHO this would require first getting the spare laser up to spec. or temporarily swapping in the spare then replacing it with the rebuilt laser. Note that LTW will not align the laser with the phase-correcting EOM (PCEOM) for the FSS servo installed between the oscillator and the amplifier. 6 About 6 years ago, P. King and R. Savage made detailed comparisons with recently calibrated (at NIST) Scientech power meters. 7 http://blue.ligowa.caltech.edu/ilog/pub/ilog.cgi?group=detector&date_to_view=05/20/2001&anchor_to_scroll_to=2001:05:20: 19:26:49-rick 8 http://blue.ligowa.caltech.edu/ilog/pub/ilog.cgi?group=detector&date_to_view=09/15/2003&anchor_to_scroll_to=2003:09:16: 19:54:33-rick Page 9 of 21

4.1.2 Swap out laser in the LVEA Swapping out the laser will require removing the PCEOM from the existing laser and installing it into the replacement laser, modematching the new laser to both the PMC and reference cavity paths, and adjusting the frequency stabilization servo (FSS) and intensity stabilization servo (ISS) to work with the new actuators. Effort and Cost Swapping out the laser, connecting the cooling lines and electrical cables, leveling the laser, etc. should take less than one day. It is the ancillary tasks, detailed below, that will consume more time. 4.1.3 Install PCEOM outside of laser head Historically, LTW has refused to align the 126MOPA lasers with the PCEOM installed between the oscillator and the amplifier. Jim Kayser from LTW indicates that one of the modematching lenses (L3) must be replaced when a PCEOM is installed. Experience with the LHO lasers indicates that output power is reduced and beam quality compromised when the PCEOM is installed. One option is to locate the PCEOM outside the laser box in the path of the full output power beam. Originally, the PCEOM was installed between the MO and the PA to avoid exposure to the amplified beam. However, UF measurements indicated that the NewFocus EOMs can withstand the full output power of the laser. A 3 month long-term exposure test at LHO did not indicate any measurable degradation in transmission or transmitted beam quality (using a BeamView camera) with the PCEOM transmitting the full 11 W laser output. Effort and Cost Diplicate the UF goniometer mounts for the IOO EOMs. Cost ~$1k per laser. Installation effort ~1 hour, significantly less than the effort to install the PCEOM inside the laser head. 4.1.4 Reconfigure upstream optical layout and modematching The present PSL/IOO table layout 9 resulted from efforts to reduce the number of optical components to the absolute minimum. At the time the layout was designed, the thinking was that we would avoid swapping out the whole laser at all costs, choosing to replace amplifier modules if necessary while preserving the MO and associated servo actuators, PCEOM, etc. Maser oscillators have been replaced several times, but the recent LLO swap is the first time that an amplifier has been replaced and they chose to replace the whole laser. With this scenario in mind, we propose revising the optical layout as shown in Figure 7. 9 http://www.ligo.caltech.edu/docs/d/d000336-00.pdf Page 10 of 21

Figure 7 Revised upstream optical layout to facilitate swapping lasers and maintaining constant power to the FFS and PMC paths.. The laser would be moved back about two feet. A half-wave plate and polarizer would be installed at the laser output. This would enable maintaining a constant power level to the inputs of both the PMC path and the FSS path. Next,. We would mount the PCEOM then a modematching telescope (MMT) between the laser power adjustment polarizer and the splitter that samples the beam for the FSS path. The polarizer would help to reduce RFAM by cleaning up the polarization before the EOM. The MMT would match the beam to the PMC so that the presently-installed two-lens MMT would be removed. The two-lens MMT between the pick of and the frequency shifter in the FSS path would be replaced with different lenses to match the beam to the AOM. Once the layout is upgraded, replacement of the laser will simply require directing a sample of the beam downstream of the MMT to an analyzer and adjusting the MMT lenses to give the spot size and beam location required for the PMC. Once the beam is aligned into the PMC it should be sufficiently aligned and modematched to the FSS path. Effort and cost We now have a pretty good selection of CVI lenses (at least at LHO) but we will likely need a couple that we don t have. Approx. $1 k per laser should be enough for the lenses and mirrors. The estimated effort for the first swap is 1-2 days, slightly longer than just remodematching to both the PMC and FSS paths. Subsequent exchanges should be able to be completed in a significantly shorter time, ~1/2 day. 4.1.5 PSL servo optimization Assuming that the whole laser has been replaced, both the FSS and ISS servos will have new actuators. The actuation coefficient for the NPRO FAST actuator will likely vary by about 25% and may have the opposite sign. The FAST/PC crossover frequency and the loop UGF Page 11 of 21

will have to be measured and adjusted. If the current shunt for the replacement laser has not been upgraded, it will have to be swapped. The ISS loop shape will be measured and UGF adjusted if necessary. Effort and Cost The estimate time required to measure and tune both loops is about 1 day. No associated costs are expected. 4.2 H2 126 MOPA This laser has been operating continuously since December 1998, more than 40,000 hours. The performance of the H1 laser was last assessed on June 13, 2003 10. At that time no effort was made to assess the maximum power that could be delivered by the laser, i.e. the current was not increased and no effort was made to align components inside the laser head. The laser output power was 6.6 watts. Apparently the beam quality was particularly good because the PMC (old style) was 95% indicating that at least 95% of the output power was in the TEM 00 mode. Options for improvement 4.2.1 Rebuild 126MOPA at LTW Assuming that the laser output power cannot be increased by a factor of two by alignment, temperature optimization, and MO and PA current increases, the laser would have to be sent back to LTW for rebuilding. Effort and Cost The rebuild and laser replacement effort and cost would be identical to that for the H1 PSL described in Section. 4.1, above. 4.3 LHO Spare 126 MOPA The spare 126MOPA laser at LHO is presently being evaluated in the optics lab. This laser has likely logged only a few thousand hours, mostly for the long-term exposure testing of the EOM outside the laser box and for a SURF project or two. The output power is about 7 watts with a reasonable, but not spectacular, looking beam. D. Cook has been working on alignment of the laser for the past few weeks. Based on that experience and suggestions from Jim Kayser at LTW, the amplifier head was opened and visually inspected. It appears that the last two (downstream) 20-W diode bars are not emitting as much light as the other six. This may be responsible for the difficulties encountered with achieving high power with good beam quality. Options for improvement 4.3.1 Replace two diode bars ourselves If further efforts to improve performance are not successful within the next few weeks, we will have to make a decision about the amplifier. Rather than send it back to LTW for a $30k rebuild, we might investigate trying to replace (with guidance from LTW) the last two diode 10 http://blue.ligowa.caltech.edu/ilog/pub/ilog.cgi?group=detector&date_to_view=06/13/2003&anchor_to_scroll_to=2003:06:13: 10:06:08-rick Page 12 of 21

bars in the amplifier. Purchasing two bars from LTW and doing the work ourselves should save a lot of money and give us valuable experience. We may need to make or purchase some fixtures and diagnostic tools, but our clean facilities should be at least as good a what exists at LTW. Obviously this effort would require some support (advice) from LTW. Effort and Cost A WAG for the cost of the diodes is $2k each. We may need a few $k for fixturing. If we need a specialized power meter it is likely to be ~$2k. Total ~$10k. Time required for coordinating with LTW, designing fixtures, setting up the work area, replacing diodes, realigning and peaking ~ 10 man days. 4.4 L1 126 MOPA On June 26, 2003 the total laser output power was 4.8 W. This laser was subsequently removed and replaced with the LLO spare laser. The present full output power of the laser is 7.5 watts. Increasing the amplifier diode current and optimizing the alignment may increase the power somewhat, but recovering the full 11 W specified output is less likely. 4.5 LLO Spare 126 MOPA This laser is presently at LTW to have the amplifier rebuilt. We understand that the estimated cost is about $30k and that the scope of the work includes replacing the pump diodes in the amplifier, but not in the NPRO. Assessing the performance for of this laser after the LTW rebuild will help determine what course of action to take with our other lasers that are producing less than the specified output power. 5 IOO EOMs Presently each interferometer utilizes three NewFocus resonant EOMs (Model 4003) with custom V-coat AR coatings. These three modulators are mounted in series on the PSL/IOO table. A single beam waist is formed near the center of the middle modulator. The modulator chain is preceded and followed by Brewster polarizers and the automated rotatable half-wave plate that controls the power into the MC. 5.1 H1 EOMs The most recent measurement of the transmission of the H1 EOM chain (from input to first EOM to output of top PSL/IOO periscope mirror) were performed on Aug. 27, 2003 11. The measured transmission was 87%. Note that the 24.5 MHz EOM was replaced several months earlier because it was producing a lot of scattered light and the transmission was down. While visual inspection immediately after removing it revealed apparent optical damage, subsequent evaluation by NewFocus and by the UF group did not reveal any contamination or optical damage. This remains a mystery, with the only plausible (?) explanation that the crystal is somehow healing itself even at room temperature. 11 http://blue.ligowa.caltech.edu/ilog/pub/ilog.cgi?group=detector&date_to_view=08/27/2003&anchor_to_scroll_to=2003:08:27: 14:39:18-rick Page 13 of 21

Options for improvement 5.1.1 Implement single-eom modulation Implement the single-modulator technique investigated by Dick Gustafson and Paul Schwinberg s SURF student, Lucas Koerner during the summer of 2002 12. Effort and Cost Design and fabricate modulation summation electronics ~1 man week. If we adopt this scheme, we will have lots of spare resonant EOMs. The broadband EOMS (with V-coat) cost ~$5k.ea. new. However, converting resonant EOMs to broadband models should be less than $2k ea. Implementation and assessment ~ 2 man days Ifo. performance assessment ~ 2 days. 5.2 H2 EOMs The transmission of the H2 EOM chain was measured to be 75% on June 13. 2003 5. The cost and effort to implement the single EOM scheme would be the same as for H1 (refer to Section 5.1.1. 5.3 LHO Spare EOMs We have a full set of spare resonant EOMs (I think, we should check this). If we go with the single-modulator scheme, we will have to convert one or more resonant modulators to broadband models for spares. 5.4 L1 EOMs 5.5 LLO Spare EOMs 6 Modecleaner The modecleaner transmissions are inferred from measurements made at the output of the top periscope mirror on the PSL/IOO table and at the REFL port and AS port tables. Because we expect that the Faraday Isolator is the principal source of losses for the optics downstream fo the MC, all of the downstream losses are attributed to the FI. We suspect that this is the case with the H2 FI, where the losses are particularly high. This may not be true for H1. 6.1 H1 MC The most recent estimate of the H1 MC transmission, recorded in the detector elog on Dec. 31, 2003 13, is 64%. The most recent measurement of the MC visibility that we could find, recorded on Aug. 9, 2001 14, was 86%. This is a lower limit because the locked level included the modulation sidebands. Thus the H1 MC mirrors appear to have much higher losses than expected. 12 LIGO-T020208-00-W 13 http://blue.ligowa.caltech.edu/ilog/pub/ilog.cgi?group=detector&date_to_view=12/31/2002&anchor_to_scroll_to=2002:12:31: 08:32:34-peterF 14 http://blue.ligowa.caltech.edu/ilog/pub/ilog.cgi?group=detector&date_to_view=08/09/2001&anchor_to_scroll_to=2001:08:09: 08:03:02-luca Page 14 of 21

Options for improvement 6.1.1 Clean MC optics As cleaning SOS optics in-situ is difficult at best we would likely remove the optics from the suspension towers to drag wipe and re-hang. It may be more likely that the suspensions would be removed and replaced with new optics as it is doubtful that cleaning will resolve MC visibility issues. 6.1.2 Replace MC optics Should the modecleaner optics need replacing or cleaning, the installation should be coordinated with the Faraday isolator replacement, if required. To exchange the MC components in HAM1(HAM7 for H2) and HAM2 (HAM8 for H2) would require approximately three to five days of pre-vent preparations. Preparations include writing a procedure and receiving approvals, placing the IFO to standard configuration and freezing,. pulling IOT1 (IOT7 for H2) and probably ISCT1 (ISCT7 for H2) tables, staging clean rooms, platforms, class B tools etc. New SOS suspensions and hardware to be installed will need to be process to class A standards for vacuum compatibility in advance. Once doors are removed the in situ work will need to be completed in 2-3 days. For every 10hrs at atmosphere it will cost us 2 to 3weeks of pump down time. Assuming that we have a spare set of MC optics, there would be no additional cost. Labor will probably include several of the LHO resident staff along with UFO personnel. An approximation of the labor costs to reprocess the SOS suspensions ~ 300 man hours of LHO staff time. A guess of the labor costs would be ~ 150 man hours of LHO to complete the in situ work. 6.2 H2 MC The H2 MC transmission was noted in the elog on May 2, 2003 15. The estimate of the transmission inferred from power measurements made outside the vacuum envelope was 88.5%, significantly higher than for H1 (64%). The H2 MC visibility was measured to be 97% on Aug. 16, 2001 16. 6.3 LHO Spare MC optics Hopefully the optics that are removed could be cleaned and re-processed to meet specifications. If this is not the case, a significant effort and cost would be required to procure, polish, and coat the optics. We will not estimate the cost here. 15 http://blue.ligowa.caltech.edu/ilog/pub/ilog.cgi?group=detector&date_to_view=05/02/2003&anchor_to_scroll_to=2003:05:02: 00:22:51-vagesh 16 http://blue.ligowa.caltech.edu/ilog/pub/ilog.cgi?group=detector&date_to_view=08/16/2001&anchor_to_scroll_to=2001:08:16: 16:26:56-ottaway_d Page 15 of 21

6.4 L1 MC 6.5 LLO Spare MC optics 7 Faraday Isolator Because of how the measurements are made, we can only estimate the losses of the whole group optics between the MC and the RM, which includes the FI. We attribute all of the downstream optics losses to the FI. The FI is currently installed utilize thin-film polarizers that are known to be more lossy than Brewster angle polarizers. Also, there has been evidence of clipping at the FI aperture in the past. 7.1 H1 FI The Dec. 31, 2003 estimate of the transmission of the optics downstream of the MC 13 is 92.8%. This is not out of line with what is expected for the losses from the eight surfaces of the FI, which includes a half-wave plate. 7.2 H2 FI The May 2, 2003 estimate of the losses of the optics downstream of the H2 MC 15 was 81.7%. This is significantly higher than the estimate of 92.8% for H1. There have been a number of attempts to explain power variations on the H2 REFL table (ISCT7), and temperatureinduced changes in the FI optics have been suspected. Options for improvement 7.2.1 Replace FI The UF group has been working on a larger aperture FI that utilizes Brewster polarizers. To exchange the Faraday isolator and components in HAM7 (H2) would require approximately three days of pre-vent preparations. Preparations include writing a procedure and receiving approvals, placing the IFO into standard configuration and freezing, pulling IOT7 and probably ISCT7 tables, staging clean rooms, platforms, class B tools, etc. Hardware to be installed will need to be processed to class A standards for vacuum compatibility in advance. Once doors are removed the in-situ work will need to be completed in 1-2 days. For every 10 hrs. at atmosphere it will cost us 2 to 3weeks of pump down time. Labor will probably include several of the LHO resident staff along with UFO personnel. A guess of the labor costs would be ~ 70-100 man hours of LHO and 20 hrs of UFO time. 7.3 LHO Spare FI Obtain cost estimate from the UF group. Page 16 of 21

7.4 L1 FI 7.5 LLO Spare FI 8 Upgrade Scenarios 8.1 H1 interferometer Limiting ourselves to outside the vacuum upgrades, we could do the following: Rebuild 126MOPA laser Relocate PCEOM outside laser box and adjust laser position to allow modematching to PMC upstream of RC beamsplitter. Mount PMC inside a vacuum chamber Replace three IOO EOMs with a single EOM and implement the single-eom modulation scheme The expected H1 power budget after implementing these upgrades is shown in Figure 8. Modematching = 87.5% Visibility = 86 % T 126 MOPA T PMC T EOMS T MC T PSL IOO FARADAY RM 11 Watts 9.35 Watts 5.01 Watts 85 % 90 % 64 % 93 % T = 46 % Figure 8 Estimate of H1 power budget if only outside the vacuum upgrades were implemented. If we were to perform the outside the vacuum upgrades and clean or replace the H1 MC mirrors, assuming that we achieve the expected ~50 ppm losses, the expected power budget would be as shown in Figure 9 Page 17 of 21

Modematching = 87.5% Visibility = 95 % T 126 MOPA T PMC T EOMS T MC T PSL IOO FARADAY RM 11 Watts 9.35 Watts 7.04 Watts 85 % 90 % 90 % 93 % T = 64% Figure 9 Expected H1 power budget if the 'outside the vacuum' upgrades were inplemented AND the MC mirrors were cleaned or replaced. 8.2 H2 interferometer If we implemented the outside the vacuum upgrades for H2, i.e. Rebuild 126MOPA laser Relocate PCEOM outside laser box and adjust laser position to allow modematching to PMC upstream of RC beamsplitter. Mount PMC inside a vacuum chamber Replace three IOO EOMs with a single EOM and implement the single-eom modulation scheme, we expect that the power budget would be as shown in Figure 10. Page 18 of 21

Visibility = 95 % Visibility = 97 % T 126 MOPA T PMC PSL T EOMS T MC T IOO FARADAY RM 11 Watts 9.35 Watts 6.11Watts 85.0 % 90 % 88.5 % 82 % T = 56 % Figure 10 Expected H2 power budget if the 'outside the vacuum' upgrades were implemented. In addition, if we were to upgrade the H2 FI, we expect that the power budget would improve the the levels shown in Figure 11. Visibility = 95 % Visibility = 97 % T 126 MOPA T PMC PSL T EOMS T MC T IOO FARADAY RM 11 Watts 9.35 Watts 7.07 Watts 85.0 % 90 % 88.5 % 95 % T = 64 % Figure 11 Expected H2 power budget if the "outside the vacuum' upgrades were implemented AND the in-vacuum Faraday isolator were replaced. A summary of the cost and manpower estimates for the various upgrades is shown in the table below. Page 19 of 21

Cost and Manpower summary H1 Section Manpower Cost Rebuild 126 MOPA at LWE 4.1.1 1 day $30k Relocate PCEOM, mod. layout, swap laser 4.1.3 5 days $2k Mount PMC inside vacuum chamber 3.1.2 15 days $10k 17 Implement single EOM modulation scheme 5.1.1 9 days $2k Replace MC optics 6.1.2 20 days $0k 18 H2 Rebuild 126 MOPA at LWE 4.1.1 1 day $30k Relocate PCEOM, mod. layout, swap laser 4.1.3 4 days $2k Mount PMC inside vacuum chamber 3.1.2 5 days $10k 17 Implement single EOM modulation scheme 5.1.1 2 day $2k Upgrade Faraday isolator 7.2.1 12 days $0k 19 LHO Repair spare 126MOPA at LHO 4.3.1 10 days $10k 9 Options for increasing the laser power above the specified power The present power budgets for the interferometers shown in Section 2 indicate that we would not achieve the goal of 6 W at the RM even if the 126MOPA lasers were operating at the full 11 W output power. This is certainly the case with H1, even if all of the outside-the-vacuum upgrades are implemented. This may also be the case for H2, but we would be closer to the goal. The power budgets do not take aging of the 126MOPA into consideration and our experience indicates that the power drops fairly quickly (order of months) down to the 7 or 8 wattt level where it sometimes degrades much more slowly (order of years). Experience does not engender confidence that even a rebuilt 126MOPA could provide 11 W of output power for between 6 months and one year. Note that because of LTW s mis-calibrated power meter the delivered power is actually closer to 11 W than 10 W. We thus consider several options for increasing the laser power above the nominal 11 W level. 9.1 Building for 15 to 20 W with the current amplifier diodes Over the past several years the LTW technicians have indicated that the 126MOPAs are easily capable of delivering 15 to 20 watts with the present design 20. The quoted available power increased as they gained experience while filling LIGO s initial orders. They found that they were able to reduce the NPRO power and PA diode current required to deliver the 17 Does not include the cost of a replacement PMC cavity. 18 Assuming that spare MC optics are available. 19 UF to supply the first large-aperture, low-loss Faraday isolator. 20 The early conceptual design for the LIGO II pre-stabilized laser (http://www.ligo.caltech.edu/docs/t/t000036-07.pdf) was predicated on LTW upgrading the 126MOPA to 20 W output power. Page 20 of 21

specified 11 W output power. All of this information is anecdotal, but it may be worth discussing this possibility with LTW management. This is likely to be the lowest cost option, and the most conservative in terms of the likely increase in power and lifetime. 9.2 Replacing the current amplifier diodes with higher output power units Again, information is anecdotal, but we have heard of the availability of higher output power pump diodes for the amplifiers. Higher pump power would likely change the beam propagation through the amplifier so may require a complete rebuild. The cost of this kind of high-power rebuild and upgrade may only be incrementally more expensive than the $30k rebuilds using the present amplifier diodes. LTW technicians have indicated that the present 126MOPA power supply is capable of delivering the higher current that higher power diodes are likely to require. Additional chilling capacity may also be required. Higher capacity chillers have already been ordered and delivered to the sites, so only interfacing and installation costs would be required for the chillers. 9.3 Adding a second amplifier stage This is the option most likely to reliably deliver higher power, but is likely to be significantly more expensive than other options. LIGO has some experience with this option. The LIGO group at Stanford has experience operating a LIGO 126MOPA with a second amplifier stage mounted outside the 126MOPA box. Tests of the feasibility of this scheme could be carried out at Stanford where design issues such as single or double passing the last amplifier stage could be resolved. This option is likely the one that would enable the largest possible increase in output power. Options such as installing higher power pump diodes could be implemented in conjunction with this option to maximize output power if required. Page 21 of 21