GT Advanced Technologies: Is About to Bend the Solar Cost and Adoption Trend

Posted: March 5, 2014 by mattmargolis24 in My Publications
Tags: , , ,


GT Advanced Technologies left  investors with a cliff hanger (see quote below) during last week’s Q4 2013 conference call.  It was like watching your favorite TV show’s finale and having to wait until the fall to find out what happens.  I’m hoping to shed some light on what we might expect on March 14.  GT has been saying for quote sometime that “solar costs need to come down” and you know what I think they have done it.  I have to tip my hat again GT’s CEO Tom Gutierrez  and his team of scientists for their most recent invention.

While solar is not expected to contribute meaningfully to revenues over the next year, it remains an important part of our portfolio. However, we continue to believe that for the solar industry to fully reach its potential, cost must come down dramatically. To this end, we have a deployed a new technology that we expect will significantly impact the economics of producing solar cells and modules. This technology was developed and comes out of a research operation we established in the Bay Area over a year ago to focus on advancing the state of the art and the design and assembly of solar cells and modules. We look forward to talking with you about this development on our March ‘14 webcast.

CVD Background

Chemical vapor deposition (CVD) in simple terms is defined as the chemical process to produce high-purity, high-performance solid materials.   In GT’s case they use CVD reactors to manufacture electronic grade or solar grade polysilicon.   The two most common ways to produce polysilicon deposit is through a CVD reactor, using the Siemens or Rogers Heitz method, which require the assembly of a silicon rod within the reactor, that has to be mounted and connected so that electrical feeds pass through.  My simple translation, the current process involves a whole assembly process inside the reactor before pushing the green button to turn it on to grow polysilicon.

GT’s website currently lists three CVD reactor models  for sale; SDR 400,  SDR 500 and SDR 600.  All of the reactors currently use the Siemens-type (polysilicon rod) process to manufacture electronic grade or solar grade polysilicon.  The top end model SDR 600 produces greater than 600 MTA (metric tons annually) was introduced on October 25, 2012 according to a GT Press Release.  GT’s polysilcon processing system equipment, which is used to “to harvest, transport, process, and package the polysilicon rods produced in GT’s SDR series CVD reactors,” is also solely dedicated to the Siemens method of producing polysilicon rods.  I am a big fan of GT’s CEO and I wanted to share various comments he has made since 2012 on the solar industry.

Tom Guiterrez Aug 2012  Q2 Earnings Call “the industry (solar) must accelerate the development of technologies that will allow for profitability at significantly reduced total systems cost of $2 per watt at the system level or below”

Tom Guiterrez Nov 2013 Q3 Earnings Call “working on other downstream PV technologies that we believe will have a significant impact on the solar cost structure”

Tom Guiterrez  Feb 2014 Q4 Earnings Call “However, we continue to believe that for the solar industry to fully reach its potential, cost must come down dramatically. To this end, we have a deployed a new technology that we expect will significantly impact the economics of producing solar cells and modules”

Ground Breaking Technology

I’ve spent a lot of time covering all the bases to make sure that their most recently granted patent has not been deployed within GT’s current equipment, because it will change competitive landscape of solar for years to come.  GT’s most recent patent is detailed in within the method of making large surface area filaments for the production of polysilicon in a CVD reactor  patent was granted on 2/11/14.   The most recent granted CVD reactor patent was related to a previous patent that was filed on April 28, 2006.  The bottom line is that GT has been working on this for a while,  but this invention has yet to be deployed in GT’s CVD reactor equipment.  With that said, I believe that is about to change by the beginning of 2015.

From the comments above its obvious GT’s CEO has been broadcasting the need to reduce the total system costs of solar since August 2012, if not longer.  He specifically mentions that the total systems cost must come down.  Total systems cost to me is your soup to nuts.  For solar costs it would include the cost of buying equipment, producing polysilicon, cutting silicon wafers all the way through to a completed solar cell.

GT’s recently granted patent involves replacing the silicon rod method (Siemens) and instead using silicon shapes such as silicon tubes, ribbons and other large area filament shapes which increases the surface area for deposition of silicon.  In simple terms, the different silicon shapes have a larger surface area and kind of act like a magnet, the larger the surface area, the more polysilicon that can be deposited within the CVD reactor.  How much more? According to the patent the production throughput can be increased 30-40% without compromising the quality of product and without significant changes to the reactors.  Additionally, the CVD reactor itself would can be simplified; currently the system has subsystems and two power sources that various currents and voltages.  The new process would enable the complicated power supply and circuitry to be replaced with a simple system of current supply and temperature controls.  A simpler power supply and circuitry setup will also lead to a reduction in the cost of the capital equipment as well.  Below are some key highlights from the invention (note the MTA yield numbers are from 2006 and actually CVD yields have increased significantly since 2006) :

It is an object of the present invention to increase the throughput of conventional CVD reactors by incorporating silicon shapes, such as silicon tubes, ribbons, or other large surface area filament shapes of similar electrical properties, instead of the conventional solid slim rods, so that the initial surface area for deposition of silicon is increased. For example, using a tubular silicon filament of 50 mm diameter rather than the conventional slim rod, the production throughput can be increased by 30-40% without compromising the quality of the product and without significant changes to the reactors. The required change to the reactor design to use the alternative filament is so minor that it can be retrofitted to current CVD reactors quickly and at very modest cost. It can even more easily be incorporated into new reactors of the same basic design, with further cost reduction benefits.

This eliminates the requirement of a complex array of subsystems and two power sources; one power supply that can provide very high voltage and low current, and a second power supply that can sustain a very high current at relatively lower voltage plus the associated switching circuit. The two power supplies and related switching circuitry can be replaced with a simple system of current supply and temperature controls.

This change in design will result in lower capital equipment costs for new reactors of similar design, and for retrofit of existing reactors when required. This change and type of operation avoids cumbersome and time consuming start up procedures, lowers down time and increases productivity, for applications where the resulting purity level is acceptable.

For a conventional 7 mm diameter solid slim rod, the growth time is more than 70 hours and the reactor produces less than 231 metric tons per year when the down time between operating cycles was limited to 6 hours. When the solid slim rods are replaced with 50 mm OD tube filaments of the invention, the growth or CVD times are expected to be about 45 hours. Using the same 6 hour down time curve d for the calculation, 304 metric tons of polysilicon can be produced per year. As can be seen, the use of a 50 mm OD tubular starting filament in accordance with the invention yields about 30% more in throughput under the normal downtime of about 6 hours.

The Current Solar Landscape

PV Magazine  published an article over the winter regarding the falling price of silicon wafers.  Included in the PV Magazine article was some information from GT’s favorite source NPD Solarbuzz.  NPD Solarbuzz estimates that the cost to make silicon wafers might breach $0.20/watt in 2014.  The estimated silicon wafer cost of  $0.20 cents per watt was split between wafer processing costs (60% – 12 cents) and polysilicon production costs (40% – 8 cents).
Screenshot - 3_3_2014 , 11_36_23 PM

Solarbuzz indicates that the cost of polysilicon accounts for 40% of the total cost of silicon per watt.  GT’s CVD reactor invention can reduce the cost to polysilicon by 30-40% on the product side plus reduce capital expenditures, both of which will only enhance GT’s odds of landing significant future contracts.  In addition to the GT’s CVD reaction invention, GT also received a recent patent related to producing thin-film solar cells, which have several applications.    The Epitaxial growth on thin lamina focused on thin-film solar cells and GT’s ability to build a triple junction PV cell, that can be applied to power delivery, consumer electronics and elsewhere.

The other piece of the Solarbuzz chart is the cost of wafer processing,  which represents nearly 60% of the total cost per watt.    GT’s thin-film solar cell patent essentially enables the growth of a semiconductor material on a donor body, by firing GT’s Hyperion’s ion cannon which splits (exfoliates) the semiconductor material from the donor body, to form a super thin lamina (layer) of the semiconductor material.  The cost reductions from Hyperion will also have a significant impact on the cost to make silicon wafers.  As some of you know from my previous updates, TG is very confident in Hyperion, as evidenced by his February 2013 comment during the Q4 2012 Q&A.

Tom Gutierrez – “And quite honestly, I don’t see anybody that’s anywhere near competing with us right now, on next-generation technologies that we’re developing. The Hyperion, nobody’s got that”

The highlights within GT’s thin film solar cells and triple junction PV cells are listed below:

Epitaxial Growth on Thin Lamina


[0001] Sivaram et al., U.S. patent application Ser. No. 12/026,530, “Method to Form a Photovoltaic Cell Comprising a Thin Lamina,” filed Feb. 5, 2008, owned by the assignee of the present invention and hereby incorporated by reference, describes fabrication of a photovoltaic cell comprising a thin semiconductor lamina formed of non-deposited semiconductor material. Using the methods of Sivaram et al., photovoltaic cells, rather than being formed from sliced wafers, are formed of thin semiconductor laminae without wasting silicon through kerf loss or by fabrication of an unnecessarily thick cell, thus reducing cost. The same donor wafer can be reused to form multiple laminae, further reducing cost, and may be resold after exfoliation of multiple laminae for some other use.

[0020] Any number of electronic devices may be fabricated by the methods of this invention as shown in FIGS. 4 and 5. In some embodiments a triple junction PV cell may be fabricated with lamina such as a germanium lamina (FIG. 4A). In some embodiments an LED may be fabricated by methods of this invention with a gallium nitride lamina (FIG. 4B). Solid-state power devices–used in switching or amplifying large voltages and currents–are important components in communications, power delivery, and, increasingly, transportation applications. These devices may also be constructed by methods of this invention.


GT’s recently granted CVD reactor patent would allow them to significantly reduce the cost of producing polysilicon from current levels (by as much as 30-40%) as well as reduce the cost of capital equipment.  GT would be the first to buck the trend and switch from Siemens method of producing polysilicon to a new method that takes advantage of tube filaments with larger surface areas to increase CVD reactor throughput and lower cost per KG of polysilicon.  Strategically, it makes prefect sense to perfect the CVD reactor process and technology and focus on having it ready for the 2015 solar equipment upgrade cycle.  Additionally, Hyperion’s exfoliation methods to separate a silicon block into super thin wafers, can dramatically reduce the current silicon wafering costs as well as improve the overall yield.  If you factor in GT’s recent thin-film solar cell patent, partnered with Hyperion, GT is on track to produce the thinnest solar cells on the market.  They are also already considering triple junction PV cells, which is one way to advance the design and assembly of solar.

If GT can effectively bend the cost curve throughout the solar supply chain they will set themselves up as the leading solution in solar, head and shoulders above any competitors and on target to receive a significant piece of the solar spending pie in 2015 and beyond.  Courtesy of Solarbuzz, you can see that capital spending is expected to pick up significantly in 2015 and reach industry spending levels not seen since 2010.  GT averaged over $600m in annual solar revenue between PV equipment and polysilicon over 2010 and 2011.    I see no reason why GT can’t repeat or even double their annual revenue solar achieved over the course of 2010 and 2011 and bring in $1.2B+ in annual revenue ($1.00 + EPS) just from solar by 2016.  They are in a much better position heading into 2015 compared to 2010 based on the tools in their technology portfolio.   GT Advanced Technologies might be setting themselves up to bend the solar cost trend and help push solar installations to another level far sooner than currently projected beginning in 2015.  I am currently modeling Solar PV and polysilicon sales at $1.2B for 2016.  Several of the Wall Street analysts covering GTAT have all-in revenue estimates of $1.5B to $1.9B for 2016.  It’s clear the analysts don’t understand GTAT and need to be spoon fed information by the Obscure Analyst.

Screenshot - 3_5_2014 , 12_02_57 AM

GTAT’s technological advances solar industry as evidenced by recent patents; related to thin firm solar cells and methodology to grow polysilicon in a CVD reactor along with commercialization of Hyperion might enable GT Advanced Technologies to be the:

  • Highest yielding polysilicon equipment maker
  • Most cost effective polysilicon raw material equipment supplier/producer
  • Most cost effective polysilicon total solution in the marketplace
  • Most cost effective silicon wafer equipment supplier/producer
  • Most cost effective thin film solar cell producer for consumer electronics and power systems
  • Produce the thinnest and most efficient solar cells in the industry

If solar revenue can bring in over $1B a year in annual revenue by 2016 and solar sales are also expected to decline as a percentage of GT’s overall sales; then just how BIG is the total annual revenue projected to be over the coming years? $3B? $6B? 

However, we’re still of the opinion that the solar capital equipment market is not likely to show significant opportunity until 2015. As the solar industry returns to health, we expect to play an important role in helping our customers lower cost throughout the process of manufacturing solar cells. Solar will continue to be a part of our portfolio. However, as our diversification initiatives get traction, we expect that it will become a smaller percentage of our business in the coming years.

I’ll leave you with my research thesis…….

It’s amazing what you can learn by looking at the past and then realizing where you are in the present and where you will be in the future.

My review of the past indicates that we have not arrived at the future. The present is a lot grander than anyone can understand today.

In the future we will learn what we missed from the past. We will learn that clues from the past were absolutely a sign of things to come.

~The Obscure Analyst 2/28/14

Full Disclosure I am long GTAT and have no plans to buy or sell any holdings in the next 72 hours


  1. Barry Fitzgerald says:

    I will defer to your research for the number on solar potential for 2015, but I think they are low. There is a supply problem that needs to be remedied by new furnaces as well as upgrades as well as potentially Hyperion. The slack in the rope is gone. Growth like saw before in solar equipment might not happen but I would expect it to be closer to the near $2B of analysts than your numbers. But YOU have done the research so you are ahead of me on this.
    The LED market should yield as well for GT and there didn’t seem to be an upside represented here. I would think, given the costs of Gallium that there might be a play for Hyperion. What I don’t understand is if Hyperion can exfoliate substrates only or fully functional devices. It would seem that the hyperion would damage functional devices and be limited to substrates.
    As always, thanks for the great research!


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s